u.demog.berkeley.edujrw/biblio/eprints/ g... · 46 historical methods the level and trend of...
Post on 15-Aug-2020
1 Views
Preview:
TRANSCRIPT
Perfecting DataHISTORICAL METHODS, April–June 2010, Volume 43, Number 2Copyright C© Taylor & Francis Group, LLC
Decennial Life Tables for the WhitePopulation of the United States,
1790–1900J. DAVID HACKER
Department of HistoryBinghamton University
State University of New York
Abstract. In this article, the author constructs new life tables forthe white population of the United States in each decade between1790 and 1900. Drawing from several recent studies, he suggestsbest estimates of life expectancy at age 20 for each decade. Theseestimates are fitted to new standards derived from the 1900–1902rural and 1900–1902 overall death registration area life tables usinga two-parameter logit model with fixed slope. The resulting de-cennial life tables more accurately represent sex- and age-specificmortality rates while capturing known mortality trends.
Keywords: demographic history, demography, life table, mortality,nineteenth century, United States
Life tables summarize the effects of age-specific mor-tality rates on a real or synthetic cohort. In addition totheir descriptive value, life tables are an indispensable
tool for demographers, with many applications in the studyof mortality, fertility, migration, and population growth. Lifetables are often used in conjunction with indirect estimationmethods for the study of populations covered by a censusbut lacking a reliable vital registration system, such as manypopulations in developing countries and populations in thepast. Life tables, for example, can be used to estimate vi-tal rates from census age distributions and are required inown-child fertility analysis (see United Nations 1983, for adescription of commonly-used indirect methods).
Demographic historians of the nineteenth-century UnitedStates depend heavily on life tables and indirect estimationmethods. Although the federal government conducted a cen-sus every ten years, it did not implement a vital registrationsystem until the start of the twentieth century (the systemwas not complete until 1933).1 As a result, the timing andcontours of the demographic transition in the United States
Address correspondence to J. David Hacker, Binghamton Univer-sity, Department of History, PO Box 6000, Binghamton, NY 13902-6000, USA. E-mail: hacker@binghamton.edu
are less precisely known than that in nations such as Englandand Australia, which had comprehensive birth and death reg-istration by the mid–nineteenth century (Jones 1971; Woods2000). Despite this limitation, demographic historians havebeen able to estimate annual and age-specific birth rates, netmigration rates, and cohort trends in life cycle experiencesas far back as the early nineteenth century using census data,life tables, and indirect methods (Yasuba 1962; Coale andZelnik 1963; Kuznets 1965; Uhlenberg 1969; McClellandand Zeckhauser 1982; Tolnay, Graham, and Guest 1982;Ferrie 1996; Hacker n.d.)
Unfortunately, the results of these studies depend on asmall number of life tables, which suffer from limited geo-graphic coverage, limited temporal coverage, and a variety ofsource-based problems. The earliest life tables rely heavilyon data from Massachusetts, a small state in the Northeastcharacterized by much higher levels of urbanization, indus-trialization, and immigration and much lower levels of nup-tiality and fertility than the nation as a whole. Given the highshort-term variability in mortality rates that was characteris-tic of the nineteenth-century United States, it is also unclearwhether life tables based on a single year of data can be usedto represent mortality in a year other than the one for whichit was constructed. The failure of existing life tables to cap-ture suspected long-term trends in mortality is perhaps theirmost significant limitation. With just a handful of life tablescovering the entire nineteenth century, researchers have beenforced to make crude assumptions about long-term mortal-ity trends to conduct their analyses. As discussed in moredetail below, recent research indicates that earlier assump-tions of long-term mortality decline are in error. Mortalityincreased significantly in the mid-nineteenth-century UnitedStates before beginning its long-term decline.
This article constructs new life tables for the white pop-ulation of the United States in each decade between 1790and 1900. The first part of the article reviews research on
45
46 HISTORICAL METHODS
the level and trend of nineteenth-century mortality. Draw-ing from several recent studies, it suggests best estimates ofmale life expectancy at age 20 for each decade. The secondpart of the article investigates sex differentials in mortal-ity and suggests best estimates of female life expectancy atage 20 for each decade. The third part of the article reviewsresearch on the age pattern of male and female mortality.The results indicate that age-specific mortality rates in thenineteenth century did not match the two most frequentlycited standards: the west model of the Princeton regionalmodel (Coale, Demeny, and Vaughan 1983) life tables or the1900–1902 Death Registration Area (DRA) life tables forthe United States. It concludes, however, that the life tablesconstructed for the rural part of the 1900–1902 DRA likelyapproximate the age pattern of nineteenth-century mortal-ity. Finally, the fourth part of the article fits the decennialestimates of life expectancy at age 20 to new standards de-rived from 1900–1902 rural and overall life tables using atwo-parameter logit model with fixed slope. The resultingdecennial life tables, it is argued, more accurately representsex- and age-specific mortality rates while capturing knownmortality trends.
The Level and Trend in Nineteenth-Century Mortality
Table 1 shows life expectancy and infant mortality esti-mates from selected United States life tables in the periodbetween 1789 and 1902 by year of publication (for a morecomplete listing, including life table summaries for selectedcities, see Haines 1998). The tables were constructed froma wide variety of sources, including local bills of mortality,state and national death registration data, census data, familygenealogies, and biographical data on special populationssuch as legislators and college graduates. Table 1 also showsthe sex mortality differential at age 20, defined as the femalelife expectancy at age 20 minus male life expectancy atage 20.
Edward Wigglesworth (1793) constructed the first UnitedStates life table using Bills of Mortality for 35 New Eng-land towns in the late eighteenth century. Ezekiel B. Elliott(1858), John S. Billings (1885), Samuel W. Abbott (1898),and James W. Glover (1921) relied on death registration datain Massachusetts—the first state to implement a death regis-tration system—to calculate life tables for selected years inthe late nineteenth century.2 Levi Meech (1898) constructedthe first national life table. The lack of national death regis-tration data forced Meech to rely on an indirect approach. Heestimated cohort declines from the 1830–60 federal censuses,made adjustments from immigration data to account for thelack of a closed population and used retrospective mortalitydata published by the 1860 census to establish the age patternof death (1898, 255–59). The creation of a national death reg-istration area (DRA) in 1900 greatly facilitated the creationof life tables. Glover’s (1921) 1900–1902 DRA life tables re-lied on registration data from the 10 states and the District ofColumbia that comprised the nation’s original DRA. These
tables have been widely used by researchers to represent thelevel and age pattern of mortality in the United States at theturn of the twentieth century.
Two studies conducted in the mid-twentieth century havebeen widely cited as representative of nineteenth-centurymortality. Abram J. Jaffe and William I. Lourie Jr. (1942)relied on death registration data collected by 44 New Eng-land towns, several midsized cities, and a few larger citiesto construct life tables for the period 1826–35. The resultsindicated large differentials in life expectancy between ruralareas and large urban centers, with life expectancy at birthalmost 15 years higher in the selected towns than in the largecities of Boston, New York, and Philadelphia. Paul H. Jacob-son’s (1957) 1849–50 life tables were based on retrospectivemortality data collected by the 1850 census. Jacobson con-fined his analysis to data collected for Massachusetts andMaryland, reasoning that an arithmetic mean of their age-specific death rates would approximate those for the nationas a whole.
The life table estimates in table 1 are sorted by year of pub-lication to emphasize our relatively recent knowledge aboutnineteenth-century mortality. Researchers requiring life ta-ble data prior to the late 1970s were limited to a hand-ful of tables, which led to great uncertainty about mortal-ity trends. Inferring mortality trends in the early nineteenthcentury from existing life tables was especially problem-atic. Warren S. Thompson and Pascal K. Whelpton (1933,230–31) calculated a slow decline in the crude death ratefrom 27.8 per 1,000 in the late eighteenth century to 21.4 per1,000 in 1855 by interpolating between the Wigglesworth(1793) and Elliott (1858) life tables. Reviewing the morerecent evidence available to them in the late 1950s, ConradTaeuber and Irene B. Taeuber (1958, 269) found no con-clusive evidence of mortality decline in the first half of thenineteenth century. Yasukichi Yasuba (1962, chap. 3) sawevidence of mortality increase in the few decades preced-ing 1860 associated with increasing urbanization and declin-ing sanitary conditions. Richard Easterlin (1977), in con-trast, argued that increasing per capita income more thanoffset the negative impact of urbanization and cited life ex-pectancy estimates from the Wigglesworth (1793) and Ja-cobson (1957) life tables as evidence of significant mortalitydecline.
Most early observers agreed that the latter half of the cen-tury was characterized by substantial mortality decline; al-though opinions differed about the date of its onset. Taeuberand Taeuber (1958, 269) thought the evidence suggested an“almost continuous” decline in mortality beginning about1850. To conduct their classic study of long-term trends inwhite birth rates, Ansley Coale and Melvin Zelnik (1963)assumed a linear decline in mortality between Jacobson’s(1957) 1849–50 life tables and the 1900–1902 DRA life ta-bles. In separate analyses based on Simon Kuznets’s (1965)census-survival estimates of crude death rates, however, Ed-ward Meeker (1972) dated mortality decline after 1880, whenthe public health and sanitation movement became more
April–June 2010, Volume 43, Number 2 47
TA
BL
E1.
Lif
eTa
bles
Est
imat
esof
Infa
ntM
orta
lity
Rat
e(I
MR
)an
dL
ife
Exp
ecta
ncy
(e)
atSe
lect
edA
ges,
Uni
ted
Stat
es,1
798–
1901
Mal
eFe
mal
eD
ate
ofIn
vest
igat
or(s
)pu
blic
atio
nPo
pula
tion
Peri
odIM
Re 0
e 20
IMR
e 0e 2
0SM
D20
Wig
gles
wor
th17
93Se
lect
edM
assa
chus
etts
tow
ns17
8936
.5a
36.5
a—
Elli
ott
1858
166
Mas
sach
uset
tsto
wns
1855
0.15
5b
39.8
b39
.9b
0.15
5b
39.8
b39
.9b
Bill
ings
1886
Mas
sach
uset
ts18
78–8
20.
181
41.7
42.2
0.15
343
.542
.80.
6B
illin
gs18
86N
ewJe
rsey
1879
–80
0.15
245
.643
.30.
131
48.0
44.5
1.2
Mee
ch18
98U
.S.W
hite
s18
30–6
00.
162
41.0
40.9
0.13
442
.941
.40.
5A
bbot
t18
99M
assa
chus
etts
1893
–97
0.17
244
.141
.20.
147
46.6
42.8
1.6
Glo
ver
1921
Mas
sach
uset
ts18
900.
168
42.5
40.7
0.14
844
.542
.01.
3D
RA
stat
es19
00–0
20.
133
48.1
42.2
0.11
150
.943
.71.
6D
RA
rura
lare
as19
00–0
20.
109
54.0
45.9
0.09
055
.446
.00.
1Ja
ffe
and
Lou
rie
1942
44N
ewE
ngla
ndto
wns
1826
–35
42.9
b42
.9b
Sale
m,M
A,N
ewH
aven
,CT
1826
–36
37.8
b37
.8b
Bos
ton,
New
Yor
kC
ity,P
hila
delp
hia
1826
–37
28.0
b28
.0b
Est
imat
edU
nite
dSt
ates
41.7
b41
.7b
Jaco
bson
1957
Mas
sach
uset
tsan
dM
aryl
and
whi
tes
1849
–50
0.16
140
.440
.10.
131
43.0
41.7
1.6
Vin
ovsk
is19
72M
assa
chus
etts
1859
–61
46.4
44.0
47.3
43.0
−1.0
Kun
ze19
79G
enea
logi
es18
00–1
4—
44.7
—43
.4−1
.318
15–2
9—
44.1
—43
.3−0
.818
30–3
4—
44.4
—47
.83.
418
35–3
9—
44.6
—42
.3−2
.318
40–4
4—
43.5
—41
.7−1
.818
45–4
9—
42.5
—40
.7−1
.818
50–5
4—
42.2
—40
.1−2
.118
55–5
9—
40.8
—40
.5−0
.318
60–6
4—
40.5
—40
.2−0
.318
65–6
9—
40.1
—40
.10.
018
70–7
4—
42.6
—40
.0−2
.618
75–7
9—
43.7
—41
.5−2
.218
80–8
4—
44.4
—41
.8−2
.618
85–8
9—
43.8
—42
.2−1
.6Po
pe19
92Fa
mily
His
tori
es17
80–9
9—
47.4
—45
.6−1
.818
00–0
9—
46.4
—47
.91.
518
10–1
9—
44.6
—44
.4−0
.218
20–2
9—
43.3
—44
.91.
618
30–3
9—
44.6
—44
.60.
018
40–4
9—
41.5
—37
.1−4
.418
50–5
9—
40.8
—39
.5−1
.318
60–6
9—
41.2
—42
.21.
018
70–7
9—
44.3
—42
.2−2
.118
80–8
9—
45.8
—42
.9−2
.9(C
onti
nued
onne
xtpa
ge)
48 HISTORICAL METHODS
TA
BL
E1.
Lif
eTa
bles
Est
imat
esof
Infa
ntM
orta
lity
Rat
e(I
MR
)an
dL
ife
Exp
ecta
ncy
(e)
atSe
lect
edA
ges,
Uni
ted
Stat
es,1
798–
1901
(Con
tinue
d)
Mal
eFe
mal
eD
ate
ofIn
vest
igat
or(s
)pu
blic
atio
nPo
pula
tion
Peri
odIM
Re 0
e 20
IMR
e 0e 2
0SM
D20
Kas
akof
fan
dA
dam
s19
95N
ewE
ngla
ndfa
mili
es17
50–5
9c48
.1(s
ettle
dbe
fore
1650
)17
60–6
9c48
.317
70–7
9c43
.217
80–8
9c44
.517
90–9
9c43
.618
00–0
9c42
.018
10–1
9c40
.718
20–2
9c40
.618
30–3
9c39
.0Fe
rrie
1996
Nat
ive-
born
whi
tes
(wei
ghte
d)18
5045
.4U
rban
(wei
ghte
d)18
5038
.0R
ural
(wei
ghte
d)18
5047
.6Fo
reig
n-bo
rnw
hite
s(w
eigh
ted)
1850
35.7
Urb
an(w
eigh
ted)
1850
30.6
Rur
al(w
eigh
ted)
1850
45.3
Hac
ker
1996
Yal
egr
adua
tes
1790
–182
940
.1L
evy
1996
Mar
ylan
dle
gisl
ator
s17
50–1
764c
34.7
d
Sout
hC
arol
ina
legi
slat
ors
1750
–176
4c31
.7d
Hai
nes
1998
U.S
.whi
tepo
pula
tion
(U.S
.mod
el)
1850
0.22
838
.438
.80.
206
40.6
40.2
1.4
U.S
.whi
tepo
pula
tion
(Wes
tmod
el)
1850
0.19
538
.837
.50.
155
43.5
40.8
3.3
U.S
.whi
tepo
pula
tion
(U.S
.mod
el)
1860
0.18
843
.241
.00.
175
44.1
41.7
0.7
U.S
.whi
tepo
pula
tion
(Wes
tmod
el)
1860
0.16
543
.039
.60.
139
46.2
42.1
2.5
U.S
.whi
tepo
pula
tion
(U.S
.mod
el)
1870
0.18
544
.141
.70.
166
46.4
43.3
1.5
U.S
.whi
tepo
pula
tion
(Wes
tmod
el)
1870
0.15
644
.440
.20.
126
48.5
43.2
3.0
U.S
.whi
tepo
pula
tion
(U.S
.mod
el)
1880
0.21
440
.440
.00.
215
40.6
40.9
0.9
U.S
.whi
tepo
pula
tion
(Wes
tmod
el)
1880
0.18
040
.938
.50.
154
43.8
40.9
2.4
U.S
.whi
tepo
pula
tion
(U.S
.mod
el)
1890
0.15
746
.041
.70.
145
47.4
42.8
1.1
U.S
.whi
tepo
pula
tion
(Wes
tmod
el)
1890
0.14
845
.640
.80.
124
48.9
43.4
2.6
U.S
.whi
tepo
pula
tion
(U.S
.mod
el)
1900
0.12
848
.542
.00.
112
50.7
43.5
1.5
U.S
.whi
tepo
pula
tion
(Wes
tmod
el)
1900
0.13
547
.841
.70.
109
51.7
44.6
2.9
Hac
kere
n.d.
Prin
ceto
ngr
adua
tes
1790
–181
936
.2
Not
e.SM
D=
sex
mor
talit
ydi
ffer
entia
lata
ge20
,defi
ned
asth
efe
mal
elif
eex
pect
ancy
atag
e20
min
usm
ale
life
expe
ctan
cyat
age
20;D
RA
=de
ath
regi
stra
tion
area
.a B
oth
sexe
sco
mbi
ned.
Vin
ovsk
isre
vise
des
timat
e.b B
oth
sexe
sco
mbi
ned.
c Bir
thco
hort
s.d e 2
0es
timat
edfr
ome 2
5(M
aryl
and
legi
slat
ors=
32.2
;Sou
thC
arol
ina
legi
slat
ors=
29.6
).e E
stim
ate
mad
eby
auth
orfr
ombi
ogra
phic
alin
form
atio
npu
blis
hed
inM
cLac
hlan
1976
;Har
riso
n,M
cLac
hlan
,and
Woo
dwar
d19
80;H
arri
son
1981
;Woo
dwar
dan
dC
rave
n19
91;a
ndL
oone
yan
dW
oodw
ard
1991
.
April–June 2010, Volume 43, Number 2 49
effective, whereas Robert Higgs (1973) observed a declinein rural areas from the 1870s.
Beginning in the 1970s, new research considerably clari-fied our understanding of nineteenth-century mortality. Muchof the new research was critical of earlier studies. In a seriesof articles, Maris Vinovskis (1971, 1972, 1978) evaluated theWigglesworth (1793), Jaffe and Lourie (1942), Elliott (1858),and Jacobson (1957) life tables, all of which relied on datafrom Massachusetts. Although the Wigglesworth life tablesuggested a reasonable estimate of life expectancy at birth,Vinovskis (1971, 589) observed that Wigglesworth (1793)lacked adequate data on the age distribution of the towns,which required adjustments amounting to “little more thanintelligent guessing.” Vinovskis also noted that the townscovered by the Wigglesworth life table were not representa-tive of other New England towns in important characteristics,including their relative affluence and degree of urbanization,making it difficult to evaluate the table’s representativeness.Vinovskis (1972, 204–5) faulted Jaffe and Lourie (1942) forrelying on data from many small towns with under-registereddeaths, thus overestimating the significance of the rural-urbandifferential in mortality and understating the overall level ofmorality. Elliott (1858), Vinovskis (1972, 208–10) argued,erred in the opposite direction. To avoid including placeswith deficient record keeping, Elliott eliminated towns witha crude death rate of less than 16 per 1,000. In doing so, how-ever, Elliott likely removed towns whose true death rate waslower than 16 per 1,000 and thus overstated the true levelof mortality. Vinovskis (ibid.) also contended that Elliott’sreliance on just one year of mortality data was problematic,given the era’s high short-term variability in death rates. Fi-nally, Vinovskis demonstrated that Jacobson (1957) failed toconsider contemporary critiques of the 1850 census of mor-
tality, which noted that deaths were unevenly registered, andfailed to consider that the census was taken during a choleraepidemic, resulting in a likely overestimation of mortality de-spite the under-registration of deaths. Given these critiques,it is no surprise that Vinovskis (1978) strongly cautionedagainst inferring mortality trends from the Wigglesworth(1793), Jaffe and Lourie (1942), Elliott (1858), and Jacobson(1957) life tables. Drawing from bills of mortality and stateregistration reports, Vinovskis concluded that there was littletrend in Massachusetts mortality during the first half of thecentury.
Meech’s (1898) life table also received an extensive cri-tique. In a detailed reconstruction and analysis, Michael R.Haines and Roger C. Avery (1980) noted that Meech wasforced to make a number of assumptions to construct hislife table, including the questionable assumptions that theunderenumeration of deaths in the census and the requiredadjustment of gross to net migration were independent ofage. As a result, Haines and Avery concluded that the Meechlife table likely underestimated infant mortality and overesti-mated early childhood mortality; although it gave reasonableresults overall.
Finally, a number of researchers have cautioned againstinferring national mortality patterns from life tablesconstructed for Massachusetts and the 1900–1902 DRA(Easterlin 1977, 133; Condran and Crimmins 1979, 1;Preston and Haines 1991, 49–50; Haines and Preston 1997).Although these tables were based on relatively well-reporteddeath registration data3 and are thus reasonably accurate de-scriptions of the level of mortality and sex- and age-specificmortality patterns in those areas—they are unlikely to berepresentative of the national population. Table 2 comparesthe population of Massachusetts, the 1900–1902 DRA,
TABLE 2. Comparison of Selected Characteristics of Massachusetts and the Original Death Registration Area (DRA) Stateswith the United States as a Whole, 1850 and 1900
1850 1900
U.S. Massachusetts U.S. DRA DRA—“rural” Massachusetts
Total population 100.0 5.3 100.0 26.2 12.0 3.7Percentage urban 16.9 51.6 38.8 60.1 13.2 67.9Percentage foreign born 11.3 18.0 13.8 22.6 14.0 29.7Labor forcePercentage in agriculture 50.5 22.8 35.5 17.6 39.0 5.2Females age 20–49Average number of own children in household 2.29 1.46 1.87 1.49 1.63 1.31
Notes. The original death registration states of 1900 consisted of the six New England states (Connecticut, Maine, Massachusetts, New Hampshire,Rhode Island, and Vermont), Indiana, Michigan, New Jersey, New York, and the District of Columbia. Rural areas of the DRA were initially definedas places with less than 8,000 inhabitants. Later census definitions changed the urban/rural threshold to places of 2,500 inhabitants.Source: 1850 and 1900 IPUMS samples from Steven Ruggles, Matthew Sobek, J. Trent Alexander, Catherine A. Fitch, Ronald Goeken, Patricia Kellyet al., Integrated Public Use Microdata Series: Version 4.0 (Minneapolis, MN: Minnesota Population Center, 2009) [producer and distributor].
50 HISTORICAL METHODS
and the overall United States in 1850 and 1900 using datafrom the 1850 and 1900 Integrated Public Use MicrodataSeries (IPUMS) samples (Ruggles, Sobek, Alexander, Fitch,Goeken, Kelly et al. 2009). Massachusetts was much moreurban than the rest of the nation, had a proportionatelylarger and more rapidly growing foreign-born populationand had a much lower proportion of its labor force engagedin agriculture (the state was one of the first to industrializein the early nineteenth century). Moreover, Massachusettsenjoyed one of the best public health systems in the nationand was the leading state in the employment of women inthe labor force and in the fertility transition. Massachusettswomen age 20–49, for example, had an average of just1.5 co-residing own-children in 1850 and 1.3 in 1900,suggesting fertility rates approximately one-third lower thanthat of the nation as a whole. Thus, although Massachusettshas the best available mortality data for the nineteenthcentury, its level, trend, and age pattern of mortality areunlikely to be representative of the United States as a whole.
Table 2 also indicates that the population of the 1900–1902DRA was not representative of the nation. The initial DRAincluded the six New England states, New York, NewJersey, Michigan, Indiana, and the District of Columbia. Al-though the 1900–1902 DRA was much larger than the stateof Massachusetts—representing about 26.2 percent of thenation’s population in 1900 compared with only 3.7 percentfor Massachusetts—it varied from the rest of the nation insimilar, if less dramatic, ways. The DRA was more urbanthan the United States as a whole and its population includeda higher proportion of foreign-born residents and a lowerproportion of agricultural workers. Women in the DRA hadan average of 20 percent fewer co-residing children in thehousehold than women in the nation as a whole. It is note-worthy, however, that differences between the rural parts ofthe 1900–1902 DRA and the rest of the nation were lessextreme. Rural parts of the DRA included about the sameproportion of foreign-born residents and workers engaged inagriculture. Fertility rates in rural areas of the DRA weremuch closer to the national average.
Fortunately, just as confidence in the representativenessand accuracy of existing life tables was falling, new researchsignificantly enhanced our understanding of nineteenth-century mortality trends. Beginning with Michael Haines’sanalysis of the United States censuses of mortality (1979) andKent Kunze’s (1979) and Robert W. Fogel’s (1986) demo-graphic analyses of family genealogies, life expectancy es-timates have accumulated for each decade of the nineteenthcentury. Clayne Pope’s (1992) study of family histories isperhaps the most significant contribution for the first half ofthe century while Haines’s (1998) construction of life tablesfor the white, black, and overall populations is the most im-portant work for the last half of the century; although impor-tant research has also been published by Richard S. Meindland Alan C. Swedlund (1977); Gretchen A. Condran andEileen Crimmins (1979, 1980); Crimmins (1980); Daniel
Scott Smith (1982, 2003); Condran and Rose A. Cheney(1982); Cheney (1984); Stephen Kunitz (1984); Condran(1987); Richard Steckel (1988); Barbara J. Logue (1991);Eric Leif Davin (1993); Alice Kasakoff and John Adams(1995, 2000); Joseph Ferrie (1996, 2003); Antonio McDanieland Carlos Grushka (1995); J. David Hacker (1997); John E.Murray (1997, 2000); Chulhee Lee (1997, 2003); Susan I.Hautaniemi, Alan C. Swedlund, and Douglas L. Anderton(1999); Anderton and Susan Hautaniemi Leonard (2004);and Jeffrey K. Beemer, Anderton, and Hautaniemi Leonard(2005).
Several of the newer studies—including those by Haines(1979, 1998), Ferrie (1996, 2003), and Condran and Crim-mins (1979, 1980)—have relied on retrospective mortalitydata collected by the Census Office/Bureau of the Censusbetween 1850 and 1900. Beginning In 1850, census mar-shals were instructed to record the name of every person inthe household who died in the year prior to the census, aswell as the person’s age, sex, race, marital status, occupa-tion, and cause of death. The collected data were tabulatedand published in separate mortality volumes. These tabulateddata appear to be tailor-made for the construction of life ta-bles: the number of deaths at each age and sex can be used asthe numerator in the calculation of age-specific death rateswhile the denominator for the midyear population in each agegroup can be obtained (with some adjustment for populationgrowth in the preceding year) from the regular census enu-meration. Census officials, however, immediately discernedthat the mortality data were underreported by approximately40 percent. Life tables could only be constructed by makinglarge (and ultimately unknowable) adjustments to the num-ber of deaths reported at each age (see, e.g., Elliott 1874’s“approximate” life table for the 1870 population). Differen-tial mortality could be examined only by assuming no dif-ferentials in undercounts. Census officials clearly believed,however, that the undercount varied by region, urban/ruralresidence, and between long and recently settled states. J.D. B. De Bow (1855, 8), superintendent of the 1850 census,contended that state differentials in death rates “show not somuch in favor of or against the health of either, as they do,in all probability, a more or less perfect report of the mar-shals. Thus it is impossible to believe Mississippi a healthierState than Rhode Island.” Despite this disappointment, andthe urging of some census officials to drop the expensiveundertaking, the mortality information was deemed usefulenough to continue its collection and publication through the1900 census. More questions were added and, beginning in1880, the information was supplemented with death recordsfrom states with available registration data (Condran andCrimmins 1979).
Retrospective mortality data were undercounted for sev-eral reasons. Most obviously, solitary households left no onebehind to report the death to an enumerator. The death of ahousehold member of a larger family, especially the house-hold head, often led to the dissolution of the household.
April–June 2010, Volume 43, Number 2 51
Respondent error also led to undercounting. Deaths of in-fants and the elderly were underreported, and deaths occur-ring 6–12 months prior to the census enumeration were lesslikely to be reported than deaths occurring 0–6 months priorto the count (Condran and Crimmins 1979; Ferrie 2003). Inan early comparison of death reporting between the 1880census of mortality and the early death registration states ofMassachusetts and New Jersey, J. S. Billings (1885, xlii) ob-served that “the proportion of deaths omitted in the enumera-tors’ returns increases in a tolerably regular manner as we goback in time from the date of enumeration.” Billings calcu-lated that census undercounting of deaths in the 1880 censusincreased from about 17 percent of all registered deaths 0–6months prior to the census to 30 percent of deaths registered6–12 months prior.
Despite severe underenumeration, researchers have madecreative use of the mortality censuses. By matching deathsregistered in the DRA to deaths registered by the mortalitycensuses, Condran and Crimmins (1980, 188–90) were ableto estimate undercounts in both sources and make a more ac-curate comparison of urban and rural mortality. Ferrie (2003)used surviving original manuscript returns from the 1850 and1860 mortality census to link decedents to their household oforigin and was thus able to investigate mortality differentialsby age, occupation, wealth, nativity, migration status, andhousehold size. The use of linked microdata allowed Ferrie(1996) to make another important innovation: by relying onlyon deaths reported in the six months prior to the census, hewas able to significantly reduce respondent recall error andconstruct adult life expectancy estimates for white males byregion, urban/rural residence, and nativity. The results sug-gest a substantial advantage in life expectancy at age 20 forwhite males living in rural areas and for native-born males.
Haines (1998) has made the most significant attempt touse the mortality censuses to construct life tables. He beganby observing that the underreporting of deaths for individ-uals age 5–9, 10–14, and 15–19 appeared to be small. Byfitting age-specific mortality rates for these age groups tomodel life tables, Haines was able to avoid relying on agegroups experiencing substantial underreporting of deaths andto construct life tables for the white, black, and total popu-lations by sex for each census year between 1850 and 1900.These tables are clearly superior to their predecessors and amajor step forward in our understanding of late nineteenth-century mortality. Despite some concern about regional andtemporal differences in undercounting, mortality data werecollected for the entire nation. Thus, with the exception ofMeech’s (1898) 1830–60 life table, Haines’s tables can beconsidered the only nationally representative life tables forthe nineteenth-century United States. The availability of lifetables every 10 years between 1850 and 1900 also filledmany of the gaps between existing life tables. Contrary tomost prior assumptions, Haines’s life tables indicated thatmortality did not begin its secular decline until relatively latein the century. Life expectancy at birth was variable with-
out trend between 1850 and 1880—ranging between 38.3and 44.0 years for both sexes combined. Between 1880 and1900, however, life expectancy at birth increased from 39.4to 47.8 years (U.S. model, both sexes combined).
Researchers relying on the Haines (1998) life tables needto be aware of a few potential problems with their interpreta-tion and use. First, as Haines noted, the life tables representmortality conditions only in the year preceding each decen-nial census and thus may not be representative of the periodor decade in which they nominally represent. Haines’s 1850life table, for example, like Jacobson’s 1850 life table, mayoverstate mortality because of the 1849 cholera epidemic.Interpolating between Haines’s life tables for the intercensalperiods between 1850 and 1880 suggests that individuals liv-ing in the 1860s enjoyed the period’s lowest mortality. Theopposite is likely true. During the 1860s the United Statessuffered four years of the Civil War, a major and prolongeddepression in the postwar South, and, in 1867, another majorepidemic of cholera. The war alone is believed to have re-sulted in the death of approximately 8 percent of white menaged 13 to 43 in 1860 (Vinovskis 1989). Finally, users ofthe Haines life tables should also be aware that the shapeof age-specific mortality rates are strongly influenced by theHaines’s choice of models: model “west” of the Princetonregional model life tables and a “U. S model” derived fromthe 1900–1902 DRA life table. As discussed below, there isevidence that these models fail to accurately describe the ageprofile of mortality in the nineteenth-century United States,particularly for women in their childbearing years. Despitethese qualifications, Haines’s life tables are a major point ofreference for the latter half of the nineteenth century.4
The only studies of life expectancy prior to 1850 ap-proaching the geographic coverage of the Haines life tablesare genealogical-based estimates of adult life expectancy byKunze (1979), Fogel (1986), and Pope (1992), and meanage at death estimates by Kasakoff and Adams (1995). Be-cause genealogies observe individuals from birth to death,cohort life expectancies are easily calculated. Period esti-mates can also be made by observing deaths and years ofexposure over a given interval, typically a decade. Decenniallife expectancy estimates thus reflect mortality over the en-tire decade, not just a single year. Because individuals arefollowed over time and space, genealogical data allow theapplication of event-history methods and more sophisticatedanalyses. Kasakoff and Adams (2000), for example, wereable to examine the impact of migration on subsequent mor-tality. There are several drawbacks to the use of genealogicaldata for estimating mortality, however, including substantialunderreporting of infant and childhood deaths (thus limitingestimates to adult life expectancy), underreporting of femaledeaths, a bias toward larger and longer-lived families, lack ofcoverage of the nation’s black and foreign-born populations,small sample sizes for early birth cohorts, a bias toward mar-ried individuals who reproduce, and a bias toward familiesoriginating in the Northeast and living in the North. Kasakoff
52 HISTORICAL METHODS
38
40
42
44
46
48
50
1900189018801870186018501840183018201810180017901780
Period
Life
exp
ecta
ncy
at a
ge 2
0
Kunze
Pope
Haines (U.S. model)
Kasakoff and Adams
FIGURE 1. Male life expectancy at age 20.
and Adams’s data set, for example, was drawn from ninepublished genealogies of families whose ancestors settledin seventeenth-century New England. Although nineteenth-century descendents of the nine families can be found in allparts the nation, they were primarily located in the nation’snorthern census regions. Kunze’s (1979) and Pope’s (1992)data sets were drawn to be more representative of the re-gional distribution of the United States population. Althoughnot perfectly representative, the geographic coverage of bothsamples is reasonably representative of the overall popula-tion.
Figure 1 plots estimates of white male life expectancy atage 20 by Kunze (1979), Pope (1992), and Haines (1998),and mean age at death estimates for white males known tosurvive to age 20 by Kasakoff and Adams (1995).5 Fourobservations can be made. First, the three genealogical stud-ies report very high adult male life expectancies in the lateeighteenth and early nineteenth centuries; if the estimatesare correct, adult life expectancies in the United States atthe turn of the nineteenth century were the highest in theworld and were not again exceeded in the United States un-til circa 1920, approximately four decades after the onset ofsecular mortality decline. Second, life expectancy estimatesby Haines (1998) are about three years lower, on average,than those reported in the genealogical studies in the decadesin which they overlap and can be reliably compared. Third,although there is much variation in each study’s sources,methods, and results, it is nonetheless clear from figure 1that the genealogical-based studies support Haines’s con-
tention that mortality did not begin its secular decline untillate in the century. Finally, all three genealogy-based studiessuggest a significant increase in mortality in the antebellumera, especially in the three decades between 1830 and 1860.White male life expectancy at age 20 was approximately sixyears lower at mid-century than it was in the late eighteenthcentury.
If correct, a substantial mid-century increase in mortalityrepresents a paradox; based on an assessment of the expectedimpact of urbanization, public health, and economic growth,Easterlin (1977) had hypothesized a substantial mortality de-cline before 1880. Although urbanization increased duringthe period, facilitating the spread of infectious disease andhigher mortality, Easterlin noted that the percentage of theUnited States population living in urban areas remained mod-est until late in the century. The urban population, for exam-ple, was just 28.2 percent in 1880.6 Given an expected 10-yearurban-rural differential in life expectancy—an approximatedifferential suggested by several studies—and assuming anegligible role of public health before 1880, Easterlin esti-mated that urbanization between 1800 and 1880 reduced lifeexpectancy at birth 2.1 years, all else being equal. The nega-tive effect of urbanization, however, was more than compen-sated for by increases in the standard of living. Real nationalincome per capita increased dramatically in the period be-fore 1880, leading to significant improvements in diet andhousing.7 By assuming a theoretical relationship betweenlife expectancy and per capita income suggested by cross-sectional national data for the twentieth century (Preston
April–June 2010, Volume 43, Number 2 53
1975), Easterlin estimated that growth in real income in theperiod 1800–80 should have increased life expectancy by14 years. Together with the negative impact of urbanization,Easterlin’s model suggested that life expectancy at birth in-creased 11.9 years between 1800 and 1880.
Although a reasonable theoretical argument for decliningmortality, Easterlin conceded serious doubts in estimates ofnational income in the period before 1840, the appropri-ateness of using the relationship between income and lifeexpectancy in the twentieth century to infer the relationshipa century earlier, the possibility that public health worsenedbetween 1800 and 1880, and the need for more empiricalresearch. Given these doubts, new estimates documentinga mid-nineteenth-century mortality increase cannot be dis-missed on theoretical grounds. Moreover, indirect supportfor an “antebellum paradox” of increasing mortality dur-ing a period of strong economic growth is provided by newresearch on the anthropometric history of the nineteenth-century United States. Fogel (1986, 464–67) first called at-tention to the positive long-run correlation between cohortlife expectancy at age 10 and the final achieved heights ofwhite men. Both series decline in the early to mid–nineteenthcentury and increase late in the century. Accumulating ev-idence from other sources confirms a substantial decline inmale height for cohorts born in the mid–nineteenth century.Dora L. Costa and Richard H. Steckel (1997, 72), for ex-ample, documented a decline in stature among native-bornwhite males from a mean of 173.5 centimeters in the 1830birth cohort to 169.1 in the 1890 cohort, followed by a sub-stantial and sustained increase in heights for cohorts born inthe twentieth century. While identification of the causes ofthe decline has been difficult—hypotheses include deterio-rating diets, a worsening disease environment, the negativeimpact of early industrialization and urbanization, increas-ing rates of internal migration, and rising inequality—all re-searchers have agreed that heights declined significantly. In arecent investigation of the link between antebellum mortality,heights, and net nutrition, Michael R. Haines, Lee A. Craig,and Thomas Weiss (2003) have pointed to the importance ofan increasing nationalization and internationalization of thedisease environment. Regardless of the ultimate causes, thepositive correlation between stature and life expectancy isadditional evidence that the decline of life expectancy in themid-nineteenth century reported by recent studies reflects areal increase in mortality.
There are ample reasons to remain skeptical of the overalllevel of life expectancy reported by the genealogical studiesand the size of the suggested decline, however. Genealog-ical records suffer from two types of bias: a selection biasincurred by selecting data from demographically success-ful, native-born families, and a censoring bias incurred byexcluding individuals without complete birth and death in-formation from the analysis. Although these biases act inopposite directions—selection bias causes life expectancyestimates to be biased upwards while the censoring bias typ-
ically imparts a downwards bias—it is unlikely that theycounteract each other perfectly and consistently.8
Adult life expectancy estimates based on genealogicalsources tend to be much higher than estimates based on othertypes of sources, suggesting that selection bias dominates.Between 1785 and 1814, graduates of Yale College—an eliteNew England population with nearly complete, high-qualitydemographic data—had a life expectancy at age 20 of 40.4years; Kunze’s (1979) and Pope’s (1992) genealogical es-timates for the same period are much higher, in the mid-to-upper 40s (Hacker 1996, 121). Adult life expectancies ofother elite colonial populations were even lower than thatenjoyed by Yale graduates and were especially low in thecolonial South. Life expectancy at age 20 was 36.2 years formen graduating from Princeton College between 1709 and1819; 34.7 years for Maryland legislators born between 1750and 1764; and 31.7 years for South Carolina legislators born1750–64 (Levy 1996; Hacker 1996). Even if we assume nosignificant socioeconomic status differentials in adult mor-tality, these studies suggest that genealogical sources over-estimate male life expectancy at age 20 at the turn of thenineteenth century by 5–10 years or more. Daniel S. Levy(1996) indicates that lower life expectancy in the colonialSouth was rapidly disappearing by the late eighteenth cen-tury, however, suggesting that the overstatement of male lifeexpectancy by genealogical sources was on the lower side ofthat range, perhaps six years in the last decade of the century.
The tendency of genealogical estimates to overstate adultmale life expectancy appears to have been lower in the mid-and late nineteenth century. In the two periods where they canbe compared—1850–60 and 1870–90—Kunze’s (1979) andPopes’ (1992) combined estimates of male life expectancyat age 20 are 2.73 years higher, on average, than Haines’s(1998) estimates.9 Male life expectancy estimates derivedwith two-census methods suggest a similar differential.Table 3 shows the results of applying the Samuel Preston andNeil Bennett’s (1983) two-census method to the native-bornwhite population enumerated in the 1850 and 1860 IPUMScensuses. The method assumes the population is closed tomigration, a reasonable assumption for the native-born pop-ulation of the nineteenth-century United States. Although theresults may be biased by differential undercounting and theaccuracy of age reporting in the two censuses, the resultinglife table suggests that genealogical estimates overstate malelife expectancy at age 20 in the 1850s by about 3.5 years. Un-fortunately, substantial underenumeration of the 1870 census(see Anderson 1988, 78–82; Steckel 1991) limits comparisonto the decade 1850–60.
The lower tendency of genealogical sources to overstatelife expectancy in the mid- and late nineteenth century maybe the result of greater migration censoring in the genealogi-cal data. The opening of the trans-Appalachian west with theTreaty of Paris in 1783, the defeat of the Pan-Indian alliancein 1793, land reforms in the early nineteenth century, andthe transportation revolution of the 1830s likely increased
54 HISTORICAL METHODS
TA
BL
E3.
App
licat
ion
ofth
eP
rest
on-B
enne
ttC
ensu
s-B
ased
Met
hod
toth
eN
ativ
e-B
orn
Whi
teP
opul
atio
nby
Sex:
1850
–60
Star
tof
age
inte
rval
(x)
Popu
latio
non
June
1,18
50Po
pula
tion
onJu
ne1,
1860
Ave
age
popu
latio
nIn
terc
ensa
lgr
owth
rate
5r x
Cum
ulat
edgr
owth
rate
Stat
iona
rypo
pula
tion
inte
rval
5L
x
Stat
iona
rypo
pula
tion
abov
eag
ex,
Tx
Num
ber
surv
ivin
gto
age
xin
stat
iona
rypo
pula
tion
Est
imat
edlif
eex
pect
ancy
atag
ex,
e x
Mal
es0
1,42
3,46
22,
053,
500
1,73
8,48
10.
0366
——
——
—5
1,31
6,43
61,
698,
039
1,50
7,23
80.
0254
0.06
358
1,60
6,18
7—
——
101,
147,
038
1,44
6,00
51,
296,
522
0.02
310.
1850
31,
560,
040
14,08
0,17
731
6,62
344
.515
956,
661
1,23
3,98
41,
095,
323
0.02
540.
3064
71,
488,
128
12,52
0,13
730
4,81
741
.120
830,
860
1,05
5,63
294
3,24
60.
0239
0.42
987
1,44
9,81
811
,03
2,00
829
3,79
537
.625
654,
370
855,
794
755,
082
0.02
680.
5567
11,
317,
558
9,58
2,19
127
6,73
834
.630
548,
139
678,
327
613,
233
0.02
130.
6769
71,
206,
791
8,26
4,63
225
2,43
532
.735
452,
270
584,
639
518,
455
0.02
570.
7943
31,
147,
321
7,05
7,84
223
5,41
130
.040
372,
137
471,
681
421,
909
0.02
370.
9176
71,
056,
228
5,91
0,52
122
0,35
526
.845
310,
999
400,
900
355,
950
0.02
541.
0403
11,
007,
371
4,85
4,29
320
6,36
023
.550
256,
448
332,
500
294,
474
0.02
601.
1686
194
7,47
83,
846,
923
195,
485
19.7
5516
5,10
222
5,94
019
5,52
10.
0313
1.31
185
725,
977
2,89
9,44
416
7,34
617
.360
146,
113
194,
447
170,
280
0.02
861.
4616
073
4,39
22,
173,
467
146,
037
14.9
6593
,57
312
1,78
510
7,67
90 .
0263
1.59
881
532,
703
1,43
9,07
512
6,71
011
.470
61,01
977
,37
869
,19
90.
0237
1.72
397
387,
976
906,
372
92,06
89.
875
35,36
446
,19
440
,77
90.
0267
1.85
003
259,
355
518,
395
64,73
38.
080
20,51
524
,69
622
,60
60.
0185
1.96
310
160,
982
259,
040
42,03
46.
285
+10
,91
313
,79
812
,35
60.
0234
2.07
147
98,05
998
,05
9—
—
Fem
ales
01,
383,
318
2,02
1,27
91,
702,
299
0.03
79—
——
——
51,
266,
758
1,67
4,05
81,
470,
408
0.02
790.
0696
41,
576,
457
——
—10
1,10
6,85
61,
377,
428
1,24
2,14
20.
0219
0.19
391
1,50
7,94
313
,61
0,91
630
8,44
044
.115
968,
287
1,26
0,28
11,
114,
284
0.02
630.
3143
71,
525,
903
12,10
2,97
330
3,38
539
.920
812,
808
1,07
0,75
094
1,77
90.
0275
0.44
906
1,47
5,61
410
,57
7,07
130
0 ,15
235
.225
634,
318
791,
899
713,
109
0.02
220.
5733
31,
265,
175
9,10
1,45
727
4,07
933
.230
505,
521
638,
730
572,
126
0.02
340.
6871
81,
137,
446
7,83
6,28
224
0,26
232
.635
418,
015
531,
753
474,
884
0.02
400.
8057
21,
062,
940
6,69
8,83
522
0,03
930
.440
349,
205
447,
572
398,
389
0.02
480.
9278
31,
007,
534
5,63
5,89
520
7,04
727
.245
288,
765
355,
900
322 ,
333
0.02
091.
0420
491
3,81
44,
628,
361
192,
135
24.1
5023
8,86
830
9,83
127
4,35
00.
0260
1.15
924
874,
487
3,71
4,54
717
8,83
020
.855
161,
663
213,
946
187,
805
0.02
801.
2942
168
5,12
92,
840,
060
155,
962
18.2
6013
7,92
419
0,05
516
3,99
00.
0320
1.44
429
695,
124
2,15
4,93
113
8,02
515
.665
91,75
011
6,07
210
3,91
10.
0235
1.58
311
506,
056
1,45
9,80
712
0,11
812
.270
63,44
882
,47
672
,96
20.
0262
1.70
737
402,
343
953,
752
90,84
010
.575
36,27
548
,48
242
,37
90.
0290
1.84
534
268,
267
551,
409
67,06
18.
280
20,10
424
,19
822
,15
10.
0185
1.96
410
157,
903
283,
142
42,61
76.
685
+15
,15
416
,39
515
,77
50.
0079
2.07
183
125,
239
125,
239
——
Sour
ce:1
850–
60IP
UM
Ssa
mpl
esfr
omR
uggl
esbe
fore
the
etal
.Ste
ven
Rug
gles
,Mat
thew
Sobe
k,J.
Tre
ntA
lexa
nder
,Cat
heri
neA
.Fitc
h,R
onal
dG
oeke
n,Pa
tric
iaK
elly
etal
.,In
tegr
ated
Pub
lic
Use
Mic
roda
taSe
ries
:Ve
rsio
n4.
0(M
inne
apol
is,M
N:M
inne
sota
Popu
latio
nC
ente
r,20
09)
[pro
duce
ran
ddi
stri
buto
r].
April–June 2010, Volume 43, Number 2 55
the level and typical distance of internal migration. In theseven decades between 1790 and 1860 the area of the UnitedStates increased from 891,364 to 3,021,295 square miles andthe number of states from 16 to 33, with the greatest in-creases between 1840 and 1860 (Anderson 1988, 241, 246).The mean center of population moved further west in thetwo decades between 1840 and 1860—135.4 miles—than inany other comparable period in United States history (U.S.Bureau of the Census 1921, 34). Although we cannot be sureof the size and timing of the effect—Kunze (1979) and Pope(1992) do not report the percentage of their study popula-tions with missing death dates by decade—migrants are morelikely to be lost from observation. Without adequate attemptsto adjust the population at risk, an increase in the percentageof right-censored cases would bias life expectancy estimatesdownwards, all else being equal.10
Selection bias may also have been less important in thenineteenth century than in the eighteenth century. If selectionbias is a function of the propensity of a long-lived ancestor toproduce a large number of descendents—thus increasing theodds of producing a future genealogist—the life expectancyof earlier birth cohorts is more critical to the subsequent num-ber of descendants than that of later, larger cohorts, where wecan expect more heterogeneity. Put another way, the chancesthat a couple will produce any descendents beyond a fewgenerations is low if their mortality or the mortality of theirchildren and grandchildren is high. If mortality is low inthe first few generations, however, the chances are very highthat there will be thousands of descendants (and many po-tential genealogists) regardless of the level of mortality insubsequent generations (for a general discussion of these is-sues with regards to Chinese demographic history see Zhao2001).
Despite concerns about selection and censoring biases,it is clear from recent studies that mortality increased sig-nificantly after 1830 and remained relatively high until the1870s, at which point it began its long and sustained de-cline. Although genealogical-based estimates of male lifeexpectancy are biased upwards, especially in the eighteenthand early nineteenth centuries, they represent our best sourcefor decennial trends in life expectancy between 1790 and1890. With care, the estimates can be combined and adjustedto construct a reasonable series of adult life expectancies.
Table 4 attempts such a series by averaging the Kunze(1979) and Pope (1992) estimates of male life expectancyin each decade and adjusting the combined estimates by acorrection factor suggested by comparisons with other stud-ies. The second column shows the average Kunze and Popeestimate of male life expectancy at age 20 for each decadebetween 1790 and 1890.11 The third column shows a sug-gested correction factor for each of these decades: –6 yearsin the late eighteenth century (suggested by comparisons withthe graduates of Yale College and other special populations)and –2.73 years in the mid- to late nineteenth century (sug-gested by comparison to Haines’s [1998] life tables). The
TABLE 4. Suggested Best Estimates for Male LifeExpectancy at Age 20 (e20)
Period
Male e20 fromgenealogical-based
studies
Suggestedcorrection
factorAdjustedmale e20
1790–99 47.4 6.0 41.41800–09 45.8 –5.5 40.31810–19 44.6 –4.9 39.71820–29 44.1 –4.4 39.71830–39 43.8 –3.8 39.91840–49 42.6 –3.3 39.31850–59 41.2 –2.7 38.41860–69 40.8 –2.7 38.01870–79 43.7 –2.7 41.01880–89 45.0 –2.7 42.21890–99 n.a. n.a. 43.2 a
aInterpolated from the 1880–89 adjusted estimate and a weightedaverage of the 1900–1902 DRA and rural DRA life tables.Sources: Kent Kunze, The Effects of Age Composition and Changesin Vital Rates on Nineteenth-Century Population Estimates from NewData (Salt Lake City, UT: Department of Economics, Universityof Utah, 1979); and Clayne L. Pope, “Adult Mortality in Americabefore 1900: A View from Family Histories,” in Strategic Factors inNineteenth Century American Economic History: A Volume to HonorRobert W. Fogel, ed. Claudia Goldin and Hugh Rockoff, 267–96(Chicago: University of Chicago Press).
correction factor is interpolated between the 1790s and the1850s, corresponding to suspected trends in regional migra-tion. The adjusted male life expectancy estimates are shownin the fourth column. Because Pope’s and Kunze’s genealog-ical estimates for adult life expectancy end with the 1880–89decade, the suggested male estimate for the period 1890–99was obtained by interpolating between the 1880–89 estimateand an estimate obtained from the 1900–1902 overall andrural DRA life tables, weighted to reflect the national levelof urbanization. (The 1900–1902 DRA life tables and theirweighting to reflect national levels of urbanization is subse-quently described in more detail.)
Correction factors for the early part of the century areclearly larger and more speculative than those in the secondhalf of the nineteenth century. Indirect evidence suggests thatthey are approximately correct, however. Given the age struc-ture of the population reported in the United States censusof 1800, the adjusted estimates in the fourth column implya crude birth rate for the white population of 51.5 birthsper 1,000 inhabitants. The unadjusted estimate, however,would imply a crude birth rate of 45.6 per 1,000, whereasa 2.73-year adjustment would imply a birth rate of 47.7 per1,000. Contemporary observers and twentieth-century de-mographers have agreed that the birth rate at the turn of thenineteenth century was between 50 and 57 per 1,000, stronglysuggesting that the six-year adjustment is justified (Grabil,
56 HISTORICAL METHODS
Kiser, and Whelpton 1958, 5; McClelland and Zeckhauser1982, 71).12 Although based in part on trends in internalmigration and the known impact of migration censoring onmortality estimates, and in part on the observed bias in thegenealogical-based estimates of life expectancy compared toother sources, the linear interpolation of the adjustment factorbetween the 1790s and 1850s is also speculative. As a result,life tables constructed from these estimates will have a largermargin of error than life tables constructed from estimatesfor the latter part of the century.
The adjusted estimates shown in the fourth column suggestthat male life expectancy at age 20 declined approximatelythree years between 1790–99 and 1850–59. Male life ex-pectancy continued to decline in the 1860s, due largely to theimpact of the Civil War. Thereafter, life expectancy beganits long-term, sustained increase. It is unlikely that mortalitywas under significant human control until circa 1880, how-ever. The adjusted series suggests that male life expectancy atage 20 did not exceed its level in the late eighteenth centuryuntil the 1880s.
The suggested series indicates a more moderate declinein antebellum life expectancy than the six-year decline sug-gested by the unadjusted genealogical estimates. The de-cline is still large, however, and remains a puzzling aspect ofnineteenth-century United States demographic history. Thesuggested revisions shown here do not negate scholars’ char-acterization of the decline as an “antebellum paradox” or theneed for more research on the causes of declining health andlongevity during a period of rapid economic growth.
Sex Differentials in Adult Life Expectancy
Estimating female life expectancy at age 20 using ge-nealogical records is a major challenge. Because women ap-pear less often in public records and change their surname atmarriage, they disappear from observation more frequentlythan men. And because genealogical records do not recordwhen right-censored individuals exit observation, female es-timates of life expectancy are based on fewer cases and sub-ject to more censoring biases than male estimates.13
Difficulties determining when women entered and exitedobservation and small sample sizes in each decade likelyexplain the highly variable sex differentials in adult life ex-pectancy reported by Kunze (1979) and Pope (1992) (seetable 1). Pope reported that women experienced a 1.6-yearadvantage in life expectancy at age 20 in the 1820s and a4.4-year disadvantage in the 1840s. Kunze reported that fe-males had a 3.4-year advantage in the period 1830–34 anda 2.3-year disadvantage in 1835–39. Such rapid shifts in sexdifferentials in life expectancy are likely spurious and relatedto poor data quality.
Unfortunately, there are few studies of eighteenth- andearly nineteenth-century female life expectancy that canbe used to evaluate potential biases in Pope’s (1992) andKunze’s (1979) estimates. Female life expectancy estimates
derived using other sources and methods (e.g., estimates fromcommunity-based reconstruction studies) are also based onincomplete data and subject to substantial selection and cen-soring biases (see Hacker 1997, for a summary of existingstudies and discussion of potential biases). Life expectancyof women married to Yale graduates at age 20, for exam-ple, was five years lower in the late eighteenth than theestimates reported by Kunze (1979) and Pope (1992). Al-though the difference is approximately equal to the differ-ence observed between the genealogical estimates and thelife expectancy of Yale graduates, more than one-in-fourYale wives had an unknown date of death, rendering anassessment of bias in the genealogical estimates uncertain.Given different assumptions about the mortality experiencesof women with a missing death date, the life expectancy ofYale wives at age 20 may have been one year higher or lower(Hacker 1996, 83, 98). Much higher proportions of missingdata and margins of error characterize other late eighteenth-and early nineteenth-century estimates of female lifeexpectancy.
For the late nineteenth century, Kunze’s (1979) and Pope’s(1992) estimates of the female life expectancy can be com-pared with Haines’s (1998) estimates. The comparison indi-cates that Kunze’s (1979) and Pope’s (1992) combined esti-mates for white females at age 20 are slightly lower (–0.36years) than Haines’s estimates in the years in which they canbe reliably compared. This is in sharp contrast to the com-parison with Haines’s estimate for white males, in which thegenealogical-based estimates were substantially higher (2.73years at age 20). Given the high proportion of missing deathrecords for women in genealogies, the difficulties determin-ing when women entered and existed the at-risk population,and the highly variable sex differentials in life expectancy re-ported by Kunze and Pope, it is tempting to conclude that thisdiscrepancy is due entirely to bias in estimating female lifeexpectancy from genealogical data. Some portion of the dif-ference in the male and female comparisons with Haines’slife table estimates may be caused by Haines’s choice ofa model life table system, however. This possibility is ex-plored in the subsequent section examining age patterns innineteenth-century mortality.
Regardless of the ultimate cause, poor data quality, incon-sistent results, and the lack of an independent assessment ofpotential bias strongly suggests that determination of the levelof and trend in female life expectancy is best inferred frommale estimates. This section discusses sex differentials innineteenth-century life expectancy, suggests a best estimatefor the differential at age 20 in each decade and calculates theresulting series of female life expectancy from the adjustedmale estimates shown in table 4. The sex differential is as-sumed to be constant before 1860, after which fertility andmortality decline are assumed to have contributed to morerapid female gains in life expectancy relative to male gains(see Preston 1976, chap. 6, for a discussion of the impact ofmortality decline on sex differentials in mortality). Estimates
April–June 2010, Volume 43, Number 2 57
are made separately for the 1860s to account for excess malemortality during the Civil War.
The best estimate of the sex differential in life expectancyfor the period before 1860 and the best estimate for eachdecade after 1870 are not obvious from existing studiesof nineteenth-century U.S. mortality. Kunze’s (1979) andPope’s (1992) estimates suggest a male advantage in life ex-pectancy at age 20 while Haines’s life tables suggest a femaleadvantage. On average, the combined Pope and Kunze esti-mates of male and female life expectancies at age 20 indicatea 0.9-year male advantage before 1860. For census years1850 and 1860, Haines’s U.S. model life tables suggest anaverage female advantage in life expectancy at age 20 of 1.1years (Haines’s life tables based on Princeton model westlife tables indicate a 2.9-year female advantage).
These contrasting results persist in the postwar era.Kunze’s (1979) and Pope’s (1992) results indicate that malesenjoyed a 2.4-year advantage, on average, in the 1870s and1880s while Haines’s U.S. model life tables indicate a 1.3-year female advantage (2.7 years using model West). At thebeginning of the twentieth century, the 1900–1902 DRA lifetable shows a 1.6-year female advantage in life expectancyat age 20, which is in close agreement with Haines’s U.S.model. The close agreement is not surprising, of course;Haines’s U.S. model life tables are based on the age pattern ofmortality in the 1900–1902 DRA. The life table constructedfor the rural parts of the 1900–1902 DRA, however, showsa female advantage in life expectancy at age 20 of just 0.1years, closer to the implied sex differential in the combinedPope and Kunze estimates.
The different sex differentials in adult life expectancy ob-served in the overall and rural DRA life tables hint thatmales may have enjoyed higher adult life expectancies inthe more rural past. Such a conclusion is supported by thedemographic literature on nineteenth-century European pop-ulations.14 A recent comparative study of mortality in ruralvillages in eighteenth- and nineteenth-century Europe andAsia (the Eurasia Population and Family History Project), forexample, reports lower female lower life expectancy at age25 in three of the four European study areas. Sex differentialsin life expectancy at age 25 was –2.3-years for Sart, Belgium(a 2.3-year female disadvantage relative to males); –1.0 yearsfor Casaluidi, Italy; –2.8 years for Madregolo, Italy; and 0.7years for Scanian parishes in Sweden, for an unweighted av-erage of –1.4 years (Campbell, Lee, and Bengtsson 2004, 66).Lower female life expectancy at age 25 resulted from a re-markably consistent pattern of higher female mortality duringprime childbearing ages across study populations, suggest-ing that maternal mortality and maternal depletion played alarge role in the consistent pattern (Alter, Manfredini, andNystedt 2004). The pattern is characteristic of mortality innational populations with life expectancy below 45 and sug-gestive of higher female mortality from pulmonary tubercu-losis, other infectious diseases, and maternal causes (Preston1976, 91).
Some evidence suggests that females in rural areas ofnineteenth-century Europe suffered higher rates of infec-tious disease relative to males than females in urban ar-eas. Dominique Tabutin and Michel Willems, for example,cite evidence that excess female mortality and susceptibilityto respiratory diseases such as tuberculosis were more pro-nounced in rural areas (cited in Alter, Manfredini, and Nyst-edt 2004). Excess female mortality extended over a greaterrange of ages and was much higher in England’s 63 “healthydistricts”—mostly rural districts with crude death rates be-low 17 per 1,000—than in the 1838–54 English Life Table(Woods 2000, 187). According to Shelia Ryan Johansson,a probable reason for the higher incidence of tuberculosisamong females and higher rates of female mortality in ruralareas of Victorian England was lower nutritional status. Agri-cultural societies in the past, she observed, routinely discrimi-nated against females by reserving most food and the vast ma-jority of meat for husbands and sons. Industrialization and theability of women to participate in the paid labor force eventu-ally ended this nutritional discrimination (Johansson 1977).Higher fertility is another possible reason for higher femalemortality in rural areas. Although maternal mortality rateswere low relative to mortality rates from tuberculosis—mostnineteenth-century estimates suggest that maternal mortal-ity averaged between 5 and 10 maternal deaths per 1,000live births (Kippen 2005)—higher rates of nuptiality andmartial fertility in rural areas increased the cumulative riskof maternal mortality. Perhaps more importantly, pregnancyand lactation imposed greater nutritional demands on womenand reduced cell-mediated immunities, increasing the risk ofcontracting tuberculosis and other opportunistic infections.15
Unfortunately, with the exception of Kunze’s (1979) andPope’s (1992) studies, estimates of sex differentials in life ex-pectancy for the nineteenth-century United States are basedon highly urban, low-fertility populations such as Mas-sachusetts in the late nineteenth century, the 1900–1902DRA, or, like Haines’s (1998) life tables, are derived frommodels based on these populations. The 1850–60 Preston-Bennett life table (table 3), however, avoids this urban, low-fertility bias by relying on the national native-born whitefemale population in the 1850 and 1860 IPUMS samples.The results suggest sex differentials in life expectancy simi-lar to Kunze’s (1979) and Pope’s (1992) genealogical-basedestimates. At age 15, the sex differential in life expectancywas –1.2 years, rising to –2.3 years at age 20. The maleadvantage in life expectancy lasted until age 35. Thereafter,females enjoyed a slight advantage in expected remainingyears of life.
Together, the results from eighteenth- and nineteenth-century European populations and the results indicated by the1850–60 Preston-Bennett life tables for native-born whitessuggest that the overall average 0.9-year male advantage inlife expectancy at age 20 reported by Kunze and Pope forthe period 1780–1859 was approximately correct.16 As indi-cated by the 1900–1902 DRA life tables, however, a female
58 HISTORICAL METHODS
TABLE 5. Suggested Best Estimates for Female LifeExpectancy at Age 20
PeriodAdjustedmale e20
Suggested sexdifferential
(female-male)Suggestedfemale e20
1790–99 41.4 –0.9 40.51800–09 40.3 –0.9 39.41810–19 39.7 –0.9 38.81820–29 39.7 –0.9 38.81830–39 39.9 –0.9 39.01840–49 39.3 –0.9 38.41850–59 38.4 –0.9 37.51860–69 n.a. n.a. 38.9 a
1870–79 41.0 –0.6 40.41880–89 42.2 0.0 42.21890–99 43.2 0.6 43.8
aAverage of period estimates from 1850–59 and 1870–79. See text.Sources: Kent Kunze, The Effects of Age Composition and Changesin Vital Rates on Nineteenth-Century Population Estimates from NewData (Salt Lake City, UT: Department of Economics, University ofUtah, 1979); and Clayne L. Pope, “Adult Mortality in Americabefore 1900: A View from Family Histories,” in Strategic Factorsin Nineteenth Century American Economic History: A Volume toHonor Robert W. Fogel, ed. Claudia Goldin and Hugh Rockoff,267–96 (Chicago: University of Chicago Press).
advantage in life expectancy at age 20 had emerged bythe turn of the twentieth century. If the overall and rural1900–1902 life tables are weighted and combined to approx-imate the urban percentage of the national population, thefemale advantage in life expectancy at age 20 was 0.9 yearsin 1900.17
Table 5 suggests best estimates of female life expectancyat age 20 between 1780 and 1860 by assuming a fixed 0.9-year advantage in male life expectancy. As shown in the thirdcolumn, the sex mortality differential was assumed to shiftin favor of females in a linear fashion between the 0.9 fe-male disadvantage in life expectancy in the period before1870 and the 0.9-year advantage in female life expectancysuggested by the weighted 1900–1902 DRA life tables. Al-though somewhat speculative, the linear shift from a maleadvantage to a female advantage in life expectancy between1870 and 1900–1902 is consistent with known changes in sexmortality differentials accompanying mortality decline, theepidemiological transition, and fertility decline. The declinein pulmonary tuberculosis, in particular, likely led to morerapid declines in female mortality relative to male mortal-ity (Preston 1976, chap. 6). Because excess male mortalityduring the Civil War likely affected sex differentials in mor-tality, the female estimate of life expectancy in the period1860–69 was obtained by averaging the adjusted female lifeexpectancy in the 1850s and 1870s. Suggested best estimates
of female life expectancy at age 20 are shown in the fourthcolumn.
The Age Profile of Nineteenth-Century Mortality
Mortality varies with age in a consistent pattern, some-times characterized as a “U” or “J” shape, across a widerange of mortality levels. Mortality rates are very high ininfancy, drop rapidly in childhood, reach their lowest level inlate childhood and adolescence, and then begin to increasein a fairly regular manner with age. Because of this consis-tency, demographers have long sought to model mortality asa function of age and overall mortality. Among other uses, anaccurate model would make it possible to identify deviationsin empirical data from model patterns (suggestive of par-ticular conditions or poor data quality), to gain insight intoenvironmental and behavioral factors that may determine de-viations, and to construct life tables from poor data, partialdata, or even a single parameter (Preston, Heuveline, andGuillot 2001, 191–2). With an accurate model, for example,it would be possible to generate decennial life tables fromthe estimates of adult life expectancy suggested in tables 4and 5. Choice of model, however, involves some guessworkand is a potential source of substantial error.
Three basic approaches have been used to model the agepattern of mortality: mathematical approaches that repre-sent mortality as a function of age, tabular approaches thatshow expected patterns of age-specific mortality rates andother life table parameters at different mortality levels, anda combination of the first two approaches that uses a math-ematical function to relate mortality in a given populationto a tabulated standard population (Preston, Heuveline, andGuillot 2001, 192–201). Early attempts to describe the re-lationship between mortality and age with a single mortal-ity function were unsuccessful (see Woods 2000, 170–90for a discussion of nineteenth-century attempts to spec-ify the “laws of vitality”). For a variety of reasons, in-cluding changes in behaviors and in the leading causesof death (e.g., smoking and cancer), the age pattern ofmortality varies enough across time and space that a sim-ple mathematical model is not practical. An attempt byLarry Heligman and John H. Pollard (1980), for example,required a complex equation with eight parameters to modelthe age profile of mortality from infancy to old age.
The second approach to modeling age patterns of mortalityhas been the publication of model life table systems—setsof “model” life tables at different levels of morality. Themost popular set of model life tables, the Princeton regionalmodels, were published by Princeton demographers Ans-ley J. Coale and Paul Demeny in 1966 and revised in 1983(Coale, Demeny, and Vaughan 1966). Coale and Demenyexamined empirical data from 326 historical and contempo-rary populations. From the 192 life tables deemed reliable,Coale and Demeny identified four regional patterns, whichthey used to construct four “families” of model life tables.
April–June 2010, Volume 43, Number 2 59
In the 1983 revision, male and female life tables are shownat 25 different levels of mortality, ranging from level 1 (fe-male life expectancy at birth equal to 20 years) to level 25(female life expectancy at birth equal to 80 years) for eachof the four regional patterns. Intermediate levels are easilyobtained by interpolation. The four groups closely conformto four regions of Europe, which was the primary source ofthe life tables. The north model is based largely on life tablesfrom Scandinavian countries. It is characterized by low infantmortality and low mortality at older ages. The east model isbased on life tables from Eastern Europe and is characterizedby high infant mortality. The south model is based mostly ontables from Southern Europe and is characterized by highmortality under age 5 and above age 65 and low mortalitybetween age 40 and 60. The west model is more of a resid-ual group and is based on the largest number of life tables,including tables from Western Europe, the United States,Canada, Australia, New Zealand, and Japan. Other modellife table systems—including those created by the UnitedNations—have been created for developing countries in Asia,Africa, and Latin America, where different environments andcauses of death lead to different patterns of mortality than arefound in Coale and Demeny’s European-dominated system(United Nations 1982).
For populations with poor vital registration data, the choiceof a model life table—and thus the assumed age profile ofmortality—typically requires some guesswork. Colin Newell(1988, 165) notes that the “general, but not always helpful,rule is to use a [model life table] system which is flexibleenough to let real features and irregularities through, butwhich is sufficiently robust to be unaffected by errors inthe data.” Because the age profile of mortality is largely theresult of environmental and behavioral factors—which de-termine the distribution of causes of death and the level ofmortality—most analysts try to rely on a model life tablesystem based on data from a nearby region with a similar en-vironment, behaviors, and level of mortality. U.S. historicaldemographers tend to rely on Coale, Demeny, and Vaughn’s(1983) west model, which is based in part on historical lifetables for the United States (including the 1900–1902 DRAlife table). Robert V. Wells (1992), for example, used thewest model to infer life expectancy at birth in colonial Amer-ica from adult and child survival estimates reported in vari-ous studies. Suspecting probable underenumeration of infantdeaths in the 1900–1902 DRA, Condran and Crimmins fittedmortality rates for the one to four age group to model Westlife tables in order to estimate life expectancy in urban andrural areas of the DRA (1980, 191). Where it can be com-pared to empirical data, the west model appears to be a goodfit for the total and white populations of the early twentieth-century United States (Haines 1979, 197; Preston and Haines1991, 66). Douglas Ewbank (1987), however, found that theage mortality profile of early twentieth-century black popu-lation of the United States more closely matched the UnitedNation’s far east model life table.
TABLE 6. Implied Coale and Demeny Life TableParameters when e20 = 40 years
Variable West North East South
MaleLevel 11.71 10.78 9.32 8.83e0 43.8 41.2 38.0 38.11000q0 160.6 155.6 250.1 192.6l20 71597 66756 61927 61801
FemaleLevel 9.71 8.68 8.07 7.41e0 41.8 39.2 37.7 36.01000q0 166.7 161.7 235.7 193.0l20 67829 63075 61168 58004
Sources: A. J. Coale, P. Demeny, and B. Vaughan, Regional ModelLife Tables and Stable Populations (New York: Academic Press,1983).
Depending on the application, the choice of model can beimportant. Preston and Haines (1991) found that choice ofregional model had very little impact on indirect estimatesof child mortality in the 1900 census (1991, 64–67). Es-timating infant mortality and life expectancy at birth fromlife expectancy at age 20, however, is problematic. Table 6shows implied estimates of male and female life expectancyat birth, infant mortality rates, and the proportion of the pop-ulation surviving to age 20 when male and female life ex-pectancy is 40 years using the four Princeton regional mod-els. The implied life expectancy at birth for males rangesfrom a high of 43.8 years in model West to a low of 38.0years in the east model, a difference of nearly six years.Implied male infant mortality rates vary from a low of 156per 1,000 in the north model to a high of 250 in the eastmodel. Using the west model, nearly 72 percent of the pop-ulation survived to age 20. In the east model, the percentagewas less than 62 percent. Similar differences are observ-able for the female population. These differences illustratethe large potential error that can be incurred by relying onthe wrong model to infer a complete life table from a singleparameter.
The third approach to modeling the age profile of mor-tality, developed by William Brass (1971), uses a mathe-matical function to transform a standard life table. It thusrepresents a combination of the mathematical and tabularapproaches. Brass observed that logits of the lxs from anytwo life tables are related to one another by a linear rela-tionship, making it possible to describe a set of logits in anobserved or target population using the logits from the stan-dard table and appropriate intercept and slope values. Briefly,the logit transformation of the lx column is based on theequation:
logit(1 − lx) = Yx = 0.5 Log e(1 − lx/ lx), (1)
60 HISTORICAL METHODS
in which l0 = 1.0. The logits of an observed population,YObs(x), are related to the logits of a standard population,Ys(x), by the linear equation:
logit (Obs. lx) = YObs(x) = α + β Ys(x). (2)
To fit an observed life table to a standard table, logits of theobserved lxs are plotted against the logits of the standard lifetable. A straight line is then fitted to the points (typically withsimple linear regression or weighted regression techniques),and the intercept and slope of the line, α and β, are calculated.Once α and β are calculated, fitted logits can be computedfrom the standard logits, and the anti-logits can be taken toproduce a set of fitted lxs, as shown in the equation below:
Fitted lx = 1
1 + e2YFit (x). (3)
When the intercept (α) equals 0 and the slope (β) equals 1,the standard table will be reproduced. Values of the interceptparameter greater than 0.0 will shift the level of mortalityabove the standard table and values less than 0.0 will shiftthe level of mortality below the standard table. The slopeparameter determines the “tilt” of the table. A slope valuegreater than 1.0 indicates that infant and child mortality islower relative to adult mortality than in the standard table, anda slope less than 1.0 indicates that infant and child mortalityis higher relative to adult mortality than in the standard. Itthus becomes possible to construct a family of related lifetables from a standard life table by varying the intercept andslope parameters, calculating the anti-logits of the resultingvalues, and constructing the resulting life tables.
Although Brass (1971) suggested two sets of logits touse as a standard—a general standard and an Africanstandard—any life table can be used and logits calculateddirectly from the lx column. Appropriate choice of a standardtable can preserve variations in the age profile of mortalitythat cannot be obtained by varying the slope and interceptparameters of a standard table, such as the level of older agemortality relative to mid-age mortality or the level of infantmortality relative to childhood mortality.18 To construct hisU.S. model life tables, for example, Haines (1979) relied onthe 1900–1902 DRA life table as a standard table. With thehelp of available historical life tables from Massachusetts andother United States life tables of reasonable quality, Hainesfirst estimated the impact of urbanization and time on theslope of the age mortality profile. While more urban envi-ronments increased infant and childhood mortality relative toadult mortality, the trend in the late nineteenth century wastoward relatively lower levels of infant and child mortality(ibid., 303). From this relationship Haines determined thelikely slope parameter in each census year between 1850 and1900, effectively reducing the two-parameter logit model toa one-parameter model. The final intercept parameter was
determined by fitting the age-specific death rates of childrenage 5–19 in the mortality censuses (ibid.).
Comparison of Haines’s U.S. model life tables with thelife tables constructed using the west model as a standardindicates that the U.S. model typically yields higher infantmortality rates, lower adult mortality rates, and lower lifeexpectancy estimates at birth. In 1880, for example, Haines’sU.S. model suggests an infant mortality rate of 0.214 forwhite males and a life expectancy at birth of 40.4 years. Thelife table constructed using a west model suggests an infantmortality rate of 0.180 and a life expectancy at birth of 40.9years.
Arguably, the 1900–1902 DRA life table is a more ap-propriate standard for the nineteenth-century United Statesthan a generic standard or even model West.19 As noted intable 2, however, the DRA population was much more urbanthan the overall population in 1900, had a higher propor-tion of the population foreign born, had a lower proportionengaged in agriculture and had much lower fertility. Thecontrast is even greater with the overall population in theearly and mid-nineteenth-century United States, which wasoverwhelmingly rural and had very high fertility. Althoughvariation of the slope parameter can pick up some of the sus-pected impact of urbanization and time on the suspected ageprofile of mortality in the nineteenth century, the increase inurbanization and immigration, the decline in fertility and theagricultural sector of the economy, and the onset of the pub-lic health movement and epidemiological transition in thelater part of the nineteenth century likely affected the distri-bution of causes of death and the age profile of mortality inmore complex ways. It is likely, for example, that decliningtuberculosis in the late nineteenth century had a significantimpact on the mortality of young adults relative to infantsand older adults, especially among females. Condran andCheney (1982, 105) report that the decline in mortality frompulmonary tuberculosis explained 26.8 percent of the declinein mortality in Philadelphia between 1870 and 1900 and wasoverwhelmingly important in the decline in death rates atages 20–39.
In addition to mortality decline, rapid fertility decline inthe late nineteenth century (Hacker n.d.) likely had an impacton the age-specific mortality rates of females. Although ma-ternal mortality rates were lower than typically imagined inthe qualitative literature (Schofield 1986), repeated exposureto death in childbirth in high-fertility populations increasedfemale mortality relative to male mortality during childbear-ing age.20 Pregnancy may have been a significant risk factorin contracting tuberculosis, the leading killer of nineteenth-century Americans, and other opportunistic infections. Wecan thus expect that the shape of age-specific mortality ratesfor females in the early to mid-nineteenth century varied sig-nificantly from the shape of age-specific rates for femalesin the 1900–1902 DRA, even if the slope of the age pro-file is adjusted to account for suspected higher infant and
April–June 2010, Volume 43, Number 2 61
100
1,000
10,000
7570656055504540353025201510510Age group
Pro
port
ion
dyin
g, 1
0,00
0 q x
(lo
g sc
ale)
White females 1900-02 rural DRA,e0 = 55.4
White females 1900-02 DRA,e0 = 51.1
FIGURE 2. Proportion of dying by age group, white females in 1900–1902 death registration area.
child mortality relative to adult mortality prior to the onsetof mortality decline.
Some indication of the possible bias can be seen in figure 2,which compares the age-specific mortality rates for white fe-males in the 1900–1902 rural DRA life table with white fe-males in the 1900–1902 overall DRA life table. Age-specificrates for white females in the rural DRA were noticeablylower than that for white females in the overall DRA atmost ages, reflecting the overall higher life expectancy forfemales in rural areas. Mortality rates were roughly equalat ages 10–14, 20–24, and 25–29, however, and higher forrural females at age 15–19.21 Although we cannot be sureof the causes, higher mortality in rural areas during ado-lescence and early adulthood are suggestive of higher deathrates from tuberculosis and maternal mortality (Preston 1976;Henry 1989). Females age 20–49 residing in rural areas ofthe DRA had 9.4 percent more own children in the householdthan females in the overall DRA, increasing their exposure tomaternal mortality and risk of contracting tuberculosis andother infectious diseases.
Among the four Princeton regional models, age-specificdeath rates for white males and females in the rural andoverall 1900–1902 DRA had the closest correspondence withmodel west (after age 20, males in rural areas of the DRA hada closer relationship with the north model). Relative to themodel west level corresponding to the same life expectancyat birth, however, female death rates in the 1900–1902 DRAand rural areas of the 1900–1902 DRA were much higher inpeak childbearing years. The difference, as shown in figure 3,was especially pronounced for females residing in rural areas.
With the exception of age groups between 15 and 35, there isremarkably close correspondence between west model level15.17 and the mortality of women in the rural DRA. Age-specific death rates for rural females between ages 15 and29, however, exceeded the level expected in model West byapproximately 27 percent. The greatest divergence from themodel pattern, 36 percent, was at ages 20–24. Although asimilar pattern exists for males (not shown)—higher deathrates at ages 5–34 for white males residing in rural areasof the 1900–1902 DRA relative to the corresponding modelWest level, lower rates at ages 40 and above—the differenceswere much smaller.
Similar “humps” in age-specific mortality rates for femalesbetween the approximate ages of 15 and 45 have been ob-served in other historical populations, including eighteenth-and early nineteenth-century American populations(Rutman and Rutman 1976; Logue 1991; Hacker 1996),the mostly rural eighteenth- and nineteenth-century popu-lations studied by the Eurasia project (Alter, Manfredini, andNystedt 2004), and the mid-nineteenth-century population ofEngland (Wrigley and Schofield 1981/1989, 708–9). In theirreconstruction of English population history, for example,Edward A. Wrigley and Roger S. Schofield noted that whileage-specific mortality rates of males in the third English lifetable (1838–54) corresponded well with model North of thePrinceton regional life tables, females had higher than ex-pected rates from age 10 through age 35. The deviation fromthe model pattern prompted Wrigley and Schofield to con-struct their own model, based in part of the English life tableand in part on the north model.
62 HISTORICAL METHODS
100
1,000
10,000
0 1 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75Age group
Pro
port
ion
dyin
g, 1
0,00
0 q x
(lo
g sc
ale)
White females 1900-02 rural DRA,e0 = 55.4
Female model West level 15.17,e0 = 55.4
Female model North level 15.17
FIGURE 3. Proportion dying by age group, white females in rural areas of the 1900–1902 death registration area compared withPrinceton model West and model North. qx = proportion dying in age interval.
Did the same distinctive hump shape during childbear-ing years that characterized age-specific mortality rates forfemales in rural areas of the 1900–1902 DRA and variousEuropean and Asian populations also characterize the overallpopulation of the nineteenth-century United States? Figure 4shows the implied proportion dying in each age group fromthe Preston-Bennett 1850–60 life table shown in table 3 andthe model West level corresponding to the equivalent lifeexpectancy at age 10. Although the age pattern of mortalitysuggested by the Preston-Bennett life table is somewhat er-ratic, the distinctive deviation in age-specific mortality ratesfrom the expected pattern is again evident. For females, theimplied proportion dying in prime childbearing age groups25–29 and 30–34 exceeded the implied proportion dying inage groups 35–39, 40–44, and 45–49. Although much lesspronounced, a hump is also evident in the age-specific mor-tality pattern for white males. The two age profiles suggestthe known age and sex profiles of tuberculosis mortality.The less pronounced hump for males may also indicate theabsence of maternal mortality or different patterns of censuscoverage errors by age. Whatever the ultimate cause, the re-sults of the Preston-Bennett life table suggest that the age-sexpattern of mortality in the nineteenth-century United Statesmore closely resembled the pattern in the rural areas of the1900–1902 DRA than the pattern in the overall DRA.
Another way of approaching the question is through exam-ination of the sex mortality ratios by age. Despite higher lifeexpectancies in the rural 1900–1902 DRA than in the over-all DRA, the ratio of male-to-female mortality was lower at
most ages in the rural DRA. The difference was especiallypronounced during childbearing age.22 White females in therural DRA experienced excess mortality relative to males be-tween age groups 15–19 and 40–44. In contrast, females inthe overall DRA experienced lower mortality than males inall age groups. Among the nineteenth-century studies report-ing lower female life expectancies in early adulthood citedabove, most show excess females mortality relative to malesin prime childbearing years. Alter, Manfredini, and Nystedt,for example, report excess female mortality from age 25 to50 in six of the seven study populations in Sweden, Belgium,Italy, China, and Japan. In the rural village of Sart, Belgium,to cite a typical example, the ratio of male-to-female prob-ability of dying in the interval 25–50 was 0.78 (2004, 334).England’s third life table (1838–54) shows excess femalemortality in all five-year age groups between age 10 and 40(Wrigley and Schofield 1981/1989, 709), although the femaledisadvantage was modest. The lowest male-to-female mor-tality ratio, 0.95, was for the 25–29 age group. Excess femalemortality was much higher in England’s “healthy districts”(Woods 2000, 187), however, echoing the similar contrast be-tween sex mortality ratios in the rural and overall 1900–1902DRAs of the United States.
Although we lack death-registration data for thenineteenth-century United States, the 1860–1900 censusesof mortality allow the construction of sex differentials byage. Condran and Crimmins’s (1979) analysis of these dataindicated that, although the mortality censuses undercountedinfant and elderly deaths, the relative undercount of males
April–June 2010, Volume 43, Number 2 63
100
1,000
10,000
Age group
Pro
port
ion
dyin
g, 1
0,00
0 q x
(lo
g sc
ale)
Native-born white females
Male model west, level 9.0
Native-born white males
Female model west level
0 1 5 10 15 20 25 30 35 40 45 50 55 60 65 70
FIGURE 4. Proportion dying by age group, native-born whites in Preston-Bennett 1850–60 life table. qx = proportion dying inage interval.
and females varied little by age. Figure 5 shows the averagesex ratio in mortality in the 1860–80 censuses by age com-pared to the ratios in the overall 1900–1902 DRA and therural areas of the 1900–1902 DRA. Figure 5 also includesa plot of the average sex mortality ratios in Haines’s (1998)
1850–80 U.S. model life tables. Sex mortality ratios indicatedby the census data suggest a similar pattern to the 1900–1902rural DRA pattern: excess female mortality from adolescencethrough prime childbearing years and excess male mortalityat other ages. Sex mortality ratios in Haines’s life tables,
0.70
0.75
0.80
0.85
0.90
0.95
1.00
1.05
1.10
1.15
1.20
1.25
1.30
Age group
Rat
io m
ale
to fe
mal
e m
orta
lity
rate
1900-1902 DRA life table
Average, Haines 1850-80 life tables, U.S. model
Average, 1860-1880 U.S. mortality censuses
1900-1902 rural DRA life table
0 1 5 10 15 20 25 30 35 40 45 50 55 60 65 70 8075
FIGURE 5. Ratio of male to female probability of dying by age (qx). DRA = death registration area; qx = proportion dying inage interval.
64 HISTORICAL METHODS
however, more closely conform to the 1900–1902 overallDRA. Although Haines’s tables indicate modest excess fe-male mortality in childhood and approximately equal sex ra-tios during prime childbearing years 20–34, the age pattern ofsex mortality ratios is much closer to the overall 1900–1902DRA pattern than to the rural DRA pattern. Haines’s tablesalso suggest a lower sex differential in mortality in infancythan either the 1900–1902 overall or 1900–1902 rural DRAlife tables.
Figures 4 and 5 strongly suggest that the 1900–1902 ru-ral DRA life table is more representative of the shape ofmortality in the nineteenth-century United States than theoverall DRA life table. Age-specific mortality rates impliedby the Preston-Bennett 1850–60 life table and sex mortal-ity ratios by age in the 1860–80 censuses of mortality moreclosely conform to the pattern in the 1900–1902 rural DRAlife table than the overall DRA life table (which was itself acloser match than model West). The correspondence shouldnot be surprising: like the rural DRA life table, the populationof the nineteenth-century United States was less urban, wasmore agricultural and had higher fertility than the populationof the 1900–1902 DRA and populations used in the construc-tion of model West. Although we cannot be certain of the trueshape of age-specific mortality rates in the nineteenth cen-tury, the available evidence indicates that any model used toconstruct nineteenth-century life tables, especially life tablesfor the earlier part of the century, should draw more heav-ily from the 1900–1902 rural DRA life table than from theoverall DRA life table.
New Decennial Life Tables, 1790–1910
Two life tables constructed by Glover (1921) for the1900–1902 DRA are essential for this project: (1) the lifetable for the white population residing in the 10 DRA statesand the District of Columbia, and (2) the life table for thewhite population in the rural areas of the DRA. When thelife tables were published in 1921, the Census Bureau’s def-inition of “urban” was considered cities of 8,000 or moreinhabitants. All other places were considered rural. The Cen-sus Bureau subsequently redefined urban as places of 2,500or more inhabitants. So although nominally nonurban, the1900–1902 “rural” DRA life table is based in part on a pop-ulation residing in the modern definition of an urban area,albeit modest towns and cities of 2,500 to 8,000 inhabitants.
As shown in table 2, the population living in the 1900–1902DRA was predominately urban: over 60 percent lived in themodern definition of an urban area. Over 13 percent of thepopulation in the rural areas of the DRA also lived in anurban area. The DRA covered 26.2 percent of the national1900 population; the rural parts of the DRA only 12.0 percent.
What can be inferred about the level and pattern of nationalmortality in 1900–1902 given that nearly three-quarters ofthe population lived in states that were not part of the DRA?Although we could assume that the larger, more inclusive
life table for the overall DRA is more representative of thenational population, we know that urbanization, industrial-ization, nativity, and fertility in the DRA were not repre-sentative of the national population and likely affected theshape, level, and sex differential in mortality. A better choicemight be the 1900–1902 rural DRA life table. Although asubset of the overall DRA, the rural population was morerepresentative of the national population in terms of fertility,nativity, and occupation structure. Unsurprisingly, however,urbanization was higher in the nation as a whole than in therural areas of the DRA and was likely the most importantfactor influencing mortality.
The simplest and most defensible inference is to com-bine the overall and rural DRA life tables, using appropriateweights to produce a life table reflecting the rate of urbaniza-tion in the nation as a whole. If we assume that that nationalpopulation in 1901 was 40.2 percent urban (an interpolationof the Census Bureau’s estimate of urbanization in the na-tion as a whole in 1900 and 1910), it is a simple matter tocalculate the weight needed for each DRA life table and tocombine the two to produce one life table representative ofthe nation’s urban population.23 Relative to the overall DRAlife table, the resulting combined life table would increaseestimates of white life expectancy at age 20 by 1.5 years forwhite males and 0.9 years for white females. Sex differen-tials in life expectancy at age 20 would fall from a 1.6-yearfemale advantage in the overall DRA life table to a 0.9-yearfemale advantage in the combined table.
The combined table could in turn be used as a model forearlier years: logits of the table’s lx values could be taken andnew life tables generated by varying the slope and interceptshown in equation 2 to construct a predicted set of logits,calculating the lx values by taking the anti-logits using equa-tion 3, and constructing a new life table from the predictedlx values.
There are several problems in such an approach. Most ob-viously, urbanization was increasing rapidly in the decadesbefore the 1900 census. By design, the combined 1900–1902life table is representative of urbanization in the 1900–1902national population; nineteenth-century populations were farmore rural. Haines’s (1979) method is one possible wayaround this problem. Drawing on his analysis of availablelate nineteenth-century city and state life tables, Haines ob-served that the slope of age-specific mortality rates variedacross time and by level of urbanization in a predictable way.Haines was thus able to set the slope of his model as neededto fit the period and level of urbanization.
Although a useful innovation, Haines’s method cannot beapplied uncritically to decades early in the nineteenth cen-tury. Most of the change observed in the slope of mortalitylikely reflected the impact of public health initiatives between1880 and 1900 in the nation’s largest cities, particularly ef-forts to clean water and milk supplies.24 The net result wasfalling infant and early childhood mortality relative to adultmortality in large urban areas, despite rapidly increasing
April–June 2010, Volume 43, Number 2 65
urbanization. Because most small cities made only mod-est attempts at public health initiatives before 1900 (Duffy1990, chap. 12), it is much less certain if infant and child-hood mortality fell relative to adult mortality for the na-tion as a whole between 1850 and 1900, which Haines’smodel predicts. Indeed, as Haines (1979, 300–301) noted,the Princeton west model suggests the opposite. Betweenlevels 9 and 13—equivalent to an increase in female lifeexpectancy from 40 to 50 years and roughly spanning the in-crease in life expectancy in the late nineteenth-century UnitedStates—the west model suggests that infant and childhoodmortality should increase relative to adult mortality. Only atmortality levels above level 13 does infant and childhoodmortality begin to decline faster than adult mortality.
Given this uncertainty, a better approach would be to createa unique standard for each decade of the nineteenth centuryby repeating the weighting exercise of the 1900–1902 DRAand 1900–1902 rural DRA life tables described above, usingthe appropriate weights to yield a new standard life tablerepresentative of the urbanization level in each decade. Table7 shows the results of that exercise. Included in the tableare estimates of the mid-census level of urbanization in eachdecade (an average of the percentage urban in each of thebeginning and ending censuses), the corresponding propor-tional weights of the 1900–1902 overall and rural DRA lifetables used to create each standard, and the resulting logits ofthe tables’ lx values by age and sex. Before 1850, the nationallevel of urbanization was below that estimated in the rural1900–1902 DRA table. It was therefore assumed that the rural1900–1902 table represented the standard mortality patternfor all decades before to 1850. After 1850 urbanization be-gan to exceed the level of urbanization in the 1900–1902rural DRA life table, requiring increasing weight to be givento the overall DRA life table. The applied weighting of the1900–1902 overall DRA life table increased from 0.09 in the1850–59 decade to 0.51 in the 1890–99 decade.
From there it was a simple matter of varying the interceptin equation 2 and constructing a new life table to fit theestimates of adult life expectancy shown in tables 4 and 5.With one exception, the resulting life tables are shown intable 8. The exception is the 1860–69 life table for whitemales, which was modified to account for high mortalityamong males of military age during the Civil War. It wasconstructed in three steps. First, a “base” life table for the1860–69 period was constructed by using the average of the1850–59 and 1870–79 estimates of male life expectancy atage 20. Second, an estimate of excess male deaths in the1860–70 intercensal period was made by cohort using two-census survival methods.25 Finally, the excess male deathswere added to the base life table. Table 9 shows the results foreach year of the war.26 Unsurprisingly, mortality was highestin 1864, the last full year of the conflict. The estimates implya white male life expectancy at birth of 25.9 years, likelythe lowest level in U.S. history. Although based on crudeestimates, the method retains the unusual risk of early death
among young white males in the war. The resulting life tablefor the 1860s suggests a male life expectancy at age 20 of35.1 years, approximately two years lower than the adjustedKunze (1979) and Pope (1992) estimate. Although the baselife table and the number of excess male deaths could beadjusted to yield a perfect match, it is unclear which estimateto adjust. It is also possible that the genealogical samples,which are known to underrepresent individuals who do notmarry or reproduce, are biased against soldiers participatingand dying in the war. It was therefore decided to make nofurther adjustments to the life table.
Figures 6, 7, and 8 compare some of the new life ta-ble estimates with Haines’s (1998) life table estimates. Asshown in figure 6, the new life tables describe a decline inlife expectancy at birth from approximately 44 years in thelate eighteenth century to just over 40 years in the 1840s.Although the models assumed a slight male advantage inlife expectancy at age 20, higher male mortality in infancypushed female life expectancy at birth slightly above the maleestimate. Life expectancy at birth then declined another 3–4years in the 1850s to approximately 37 years. The decline islargely the result of the model’s prediction of increased in-fant mortality. Although the decline in adult life expectancybetween the 1840s and 1850s was relatively modest (1.4years), the model suggests that infant mortality rates rosefrom 215 to 247 per 1,000 for white males and from 190to 222 for white females. Life expectancy reached an evenlower level in the 1860s for white males—the result of theU.S. Civil War—but then increased rapidly with estimatesfor white females for the remainder of the century. Life ex-pectancy for white females increased more rapidly. By the1890s, white females enjoyed about a two-year advantage inlife expectancy at birth.
Haines’s (1998) estimates are plotted with a marker toemphasize their limitation to individual census years. In gen-eral, Haines’s life tables document a similar pattern of lowlife expectancy at midcentury and a rapid increase late inthe century. Haines’s estimates for 1860 are relatively high,however, whereas his estimate for the 1880s is relativelylow. It is difficult to know what to make of the differences.The substantial decline in life expectancy between 1870 in1880, in particular, does not correspond with known epi-demics or the qualitative literature on the mortality declinein the United States. The decline may reflect that 1880 wasa particularly unhealthy year or be an artifact of differentialcensus enumeration. The 1870 census has been long sus-pected to have undercounted the population and may wellhave undercounted mortality as well. The 1880 census, onthe other hand, benefitted from a shift from enumeration byUnited States marshals to enumeration by trained enumera-tors, a sharp increase in the number of enumerators relativeto the population, and the supplementation of mortality datain the census with available death registration data.
Figures 7 and 8 compare the proportions dying in five-year age intervals in male and female life tables selected
66 HISTORICAL METHODS
TA
BL
E7.
Stan
dard
Lif
eTa
ble
Log
itV
alue
s,Y
s(x)
,for
Dec
enni
alL
ife
Tabl
es
Dec
ade
Age
1790
–99
1800
–09
1810
–19
1820
–29
1830
–39
1840
–49
1850
–59
1860
–69
1870
–79
1880
–89
1890
–99
Mal
es0 1
−1.0
505
−1.0
505
−1.0
505
−1.0
505
−1.0
505
−1.0
505
−1.0
390
−1.0
254
−1.0
147
−1.0
029
−0.9
888
2−0
.947
6−0
.947
6−0
.947
6−0
.947
6−0
.947
6−0
.947
6−0
.934
4−0
.918
8−0
.906
6−0
.893
2−0
.877
23
−0.9
081
−0.9
081
− 0.9
081
−0.9
081
−0.9
081
−0.9
081
−0.8
941
−0.8
775
−0.8
646
−0.8
504
−0.8
334
4−0
.884
0−0
.884
0−0
.884
0−0
.884
0−0
.884
0−0
.884
0−0
.869
4−0
.852
3−0
.838
9−0
.824
3−0
.806
85
−0.8
656
−0.8
656
−0.8
656
−0.8
656
−0.8
656
−0.8
656
−0.8
508
−0.8
334
−0.8
198
−0.8
049
−0.7
871
10−0
.814
4−0
.814
4−0
.814
4−0
.814
4−0
.814
4−0
.814
4−0
.799
2−0
.781
3−0
.767
3−0
.752
1−0
.733
815
−0.7
775
−0.7
775
−0.7
775
−0.7
775
−0.7
775
−0.7
775
−0.7
629
−0. 7
457
−0.7
322
−0.7
175
−0.6
999
20−0
.722
5−0
.722
5−0
.722
5−0
.722
5−0
.722
5−0
.722
5−0
.708
8−0
.692
6−0
.679
9−0
.666
0−0
.649
425
−0.6
538
−0.6
538
−0.6
538
−0.6
538
−0.6
538
−0.6
538
−0.6
403
−0.6
246
−0.6
121
−0.5
985
−0.5
822
30−0
.588
7−0
.588
7−0
.588
7−0
.588
7−0
.588
7−0
.588
7−0
.575
0−0
.558
9−0
.546
3−0
.532
4−0
.515
835
−0.5
285
−0.5
285
−0.5
285
−0.5
285
−0.5
285
−0.5
285
−0.5
137
−0.4
964
−0.4
827
−0.4
678
−0.4
499
40−0
.467
3−0
.467
3−0
.467
3−0
.467
3−0
.467
3−0
.467
3−0
.451
2−0
.432
2−0
.417
3−0
.401
1−0
.381
645
−0.4
024
−0.4
024
−0.4
024
−0.4
024
−0.4
024
−0.4
024
−0.3
850
−0.3
645
−0.3
485
−0.3
310
−0.3
100
50−0
.329
6−0
.329
6−0
.329
6−0
.329
6−0
.329
6−0
.329
6−0
.310
9−0
.288
9−0
.271
8−0
.253
0−0
.230
555
−0.2
458
−0.2
458
−0.2
458
−0.2
458
−0.2
458
−0.2
458
−0.2
258
−0.2
023
−0.1
839
−0.1
638
−0.1
398
60−0
.137
5−0
.137
5−0
.137
5−0
.137
5−0
.137
5−0
.137
5−0
.116
3−0
.091
4−0
.071
8−0
.050
5−0
.024
965
0.00
190.
0019
0.00
190.
0019
0.00
190.
0019
0.02
370.
0494
0.06
960.
0917
0.11
8270
0.18
970.
1897
0.18
970.
1897
0.18
970.
1897
0.21
140.
2370
0.25
720.
2794
0.30
6275
0.43
290.
4329
0.43
290.
4329
0.43
290.
4329
0.45
410.
4792
0.49
910.
5211
0.54
77
(Con
tinu
edon
next
page
)
April–June 2010, Volume 43, Number 2 67
TA
BL
E7.
Stan
dard
Lif
eTa
ble
Log
itV
alue
s,Y
s(x)
,for
Dec
enni
alL
ife
Tabl
es(C
ontin
ued)
Dec
ade
Age
1790
–99
1800
–09
1810
–19
1820
–29
1830
–39
1840
–49
1850
–59
1860
–69
1870
–79
1880
–89
1890
–99
Fem
ales
0 1−1
.158
1−1
.158
1−1
.158
1−1
.158
1−1
.158
1−1
.158
1−1
.146
5−1
.132
7−1
.121
9−1
.110
1−1
.095
92
−1.0
471
−1.0
471
−1.0
471
−1.0
471
−1.0
471
−1.0
471
−1.0
336
−1.0
177
−1.0
053
−0.9
917
−0.9
755
3−1
.000
4−1
.000
4−1
.000
4−1
.000
4−1
.000
4−1
.000
4−0
.986
4−0
.969
9−0
.957
0−0
.942
9−0
.926
04
−0.9
726
−0.9
726
−0.9
726
−0.9
726
−0.9
726
−0.9
726
−0.9
582
−0.9
412
−0.9
279
−0.9
134
−0.8
961
5−0
.951
2−0
.951
2−0
.951
2−0
.951
2−0
.951
2−0
.951
2−0
.936
5−0
.919
3−0
.905
8−0
.891
1−0
.873
610
−0.8
944
−0.8
944
−0.8
944
−0.8
944
−0.8
944
−0.8
944
−0.8
794
−0.8
619
−0.8
482
−0.8
333
−0.8
154
15−0
.853
2−0
.853
2−0
.853
2−0
.853
2−0
.853
2−0
.853
2−0
.839
1−0
.822
5−0
.809
5−0
.795
3−0
.778
320
−0.7
867
−0.7
867
−0.7
867
−0.7
867
−0.7
867
−0.7
867
−0.7
742
−0.7
593
−0.7
477
−0.7
349
−0.7
196
25−0
.703
7−0
.703
7−0
.703
7−0
.703
7−0
.703
7−0
.703
7−0
.692
7−0
.679
6−0
.669
3−0
.658
0−0
.644
430
−0.6
241
−0.6
241
−0.6
241
−0.6
241
−0.6
241
−0.6
241
−0.6
138
−0.6
017
−0.5
921
−0.5
815
−0.5
687
35−0
.551
4−0
.551
4−0
.551
4−0
.551
4−0
.551
4−0
.551
4−0
.541
1−0
.528
9−0
.519
3−0
.508
7−0
.495
940
−0.4
842
−0.4
842
−0.4
842
−0.4
842
−0.4
842
−0.4
842
−0.4
735
−0.4
608
−0.4
507
−0.4
397
−0.4
264
45−0
.415
9−0
.415
9−0
.415
9−0
.415
9−0
.415
9−0
.415
9−0
.404
7−0
.391
4−0
.381
0−0
.369
4−0
.355
550
−0.3
448
−0.3
448
−0.3
448
−0.3
448
−0.3
448
−0.3
448
−0.3
328
−0.3
187
−0.3
075
−0.2
952
−0.2
804
55−0
.261
2−0
.261
2−0
.261
2−0
.261
2−0
.261
2−0
.261
2−0
.248
3−0
.233
0−0
.221
0−0
.207
8−0
.191
860
−0.1
566
−0.1
566
−0.1
566
−0.1
566
−0.1
566
−0.1
566
−0.1
426
−0.1
261
−0.1
130
−0.0
986
−0.0
813
65−0
.025
9−0
.025
9−0
.025
9−0
.025
9−0
.025
9−0
.025
9−0
.011
10.
0064
0.02
030.
0355
0.05
3970
0.14
490.
1449
0.14
490.
1449
0.14
490.
1449
0.16
050.
1791
0.19
380.
2100
0.22
9575
0.37
630.
3763
0.37
630.
3763
0.37
630.
3763
0.39
200.
4107
0.42
560.
4421
0.46
2080
+0.
6887
0.68
870.
6887
0.68
870.
6887
0.68
870.
7043
0.72
300.
7379
0.75
440.
7746
%ur
ban
5.6
6.7
7.2
8.0
9.8
13.0
17.5
22.7
26.9
31.6
37.3
1901
DR
Aw
eigh
t0.
000.
000.
000.
000.
000.
000.
090.
200.
290.
390.
5119
01ru
ralD
RA
wei
ght
1.00
1.00
1.00
1.00
1.00
1.00
0.91
0.80
0.71
0.61
0.49
68 HISTORICAL METHODS
TA
BL
E8.
New
Lif
eTa
bles
for
the
Whi
teP
opul
atio
nof
the
Uni
ted
Stat
es,1
780–
1900
Whi
tem
ales
,179
0–99
Whi
tefe
mal
es,1
790–
99
Age
q xl x
d xL
xT
xe x
Age
q xl x
d xL
xT
xe x
00.
1797
1000
0017
969
8796
144
0906
144
.10
0.16
0310
0000
1603
089
580
4418
888
44.2
10.
0394
8203
132
3580
122
4321
100
52.7
10.
0383
8397
032
1982
070
4329
308
51.6
20.
0171
7879
613
4978
081
4240
978
53.8
20.
0185
8075
114
9179
960
4247
237
52.6
30.
0110
7744
785
477
003
4162
897
53.8
30.
0117
7926
092
978
777
4167
277
52.6
40.
0087
7659
366
676
246
4085
894
53.3
40.
0094
7833
173
577
948
4088
500
52.2
50.
0253
7592
719
2137
4832
4009
648
52.8
50.
0263
7759
520
3938
2878
4010
552
51.7
100.
0195
7400
614
4536
6418
3634
815
49.1
100.
0205
7555
615
5237
3900
3627
674
48.0
150.
0309
7256
122
4335
7200
3268
397
45.0
150.
0356
7400
426
3736
3427
3253
774
44.0
200.
0419
7031
929
4934
4222
2911
197
41.4
200.
0492
7136
735
1034
8059
2890
347
40.5
250.
0434
6737
029
2432
9541
2566
974
38.1
250.
0526
6785
735
6733
0367
2542
289
37.5
300.
0435
6444
628
0231
5226
2237
433
34.7
300.
0529
6429
034
0431
2942
2211
921
34.4
350.
0475
6164
429
3130
0892
1922
207
31.2
350.
0532
6088
732
4229
6329
1898
979
31.2
400.
0541
5871
331
7728
5621
1621
316
27.6
400.
0584
5764
533
6827
9805
1602
650
27.8
450.
0652
5553
636
2226
8624
1335
695
24.1
450.
0653
5427
735
4326
2528
1322
845
24.4
500.
0806
5191
441
8524
9109
1067
071
20.6
500.
0823
5073
441
7724
3230
1060
317
20.9
550.
1122
4772
953
5622
5256
8179
6217
.155
0.11
0646
558
5147
2199
2181
7087
17.5
600.
1564
4237
366
2719
5298
5927
0614
.060
0.14
8941
411
6167
1916
3659
7166
14.4
650.
2266
3574
680
9915
8483
3974
0711
.165
0.20
8735
244
7357
1578
2740
5529
11.5
700.
3119
2764
786
2211
6679
2389
258.
670
0.29
7927
887
8307
1186
6724
7703
8.9
750.
4312
1902
582
0374
616
1222
456.
475
0.41
1119
580
8050
7777
612
9036
6.6
80+
1.00
0010
822
1082
247
629
4762
94.
480
+1.
0000
1153
011
530
5126
051
260
4.4
Whi
tem
ales
,180
0–09
Whi
tefe
mal
es,1
800–
09
Age
q xl x
d xL
xT
xe x
Age
q xl x
d xL
xT
xe x
00.
2036
1000
0020
357
8636
141
3806
141
.40
0.18
0710
0000
1807
088
254
4166
944
41.7
10.
0444
7964
335
4077
554
4051
700
50.9
10.
0430
8193
035
2379
851
4078
690
49.8
20.
0193
7610
314
6575
327
3974
146
52.2
20.
0207
7840
616
2077
547
3998
839
51.0
30.
0124
7463
892
574
157
3898
819
52.2
30.
0131
7678
610
0676
263
3921
292
51.1
40.
0098
7371
371
973
340
3824
662
51.9
40.
0105
7578
079
475
367
3845
029
50.7
50.
0283
7299
420
6535
9809
3751
323
51.4
50.
0293
7498
621
9436
9444
3769
662
50.3
100.
0218
7092
915
4535
0784
3391
514
47.8
100.
0228
7279
216
6035
9809
3400
218
46.7
150.
0344
6938
423
8434
0961
3040
730
43.8
150.
0394
7113
228
0434
8648
3040
409
42.7
200.
0464
6700
031
0932
7228
2699
769
40.3
200.
0541
6832
736
9833
2393
2691
761
39.4
250.
0478
6389
130
5431
1820
2372
541
37.1
250.
0575
6463
037
1831
3853
2359
368
36.5
300.
0477
6083
729
0129
6931
2060
721
33.9
300.
0577
6091
135
1229
5776
2045
515
33.6
350.
0519
5793
630
0728
2159
1763
791
30.4
350.
0577
5739
933
1327
8714
1749
739
30.5
400.
0588
5492
832
2926
6569
1481
631
27.0
400.
0630
5408
634
0926
1910
1471
025
27.2
450.
0704
5169
936
4224
9392
1215
062
23.5
450.
0700
5067
835
5024
4513
1209
115
23.9
500.
0865
4805
841
5722
9894
9656
7020
.150
0.08
7847
128
4139
2252
9296
4602
20.5
550.
1195
4390
052
4420
6389
7357
7616
.855
0.11
7142
989
5033
2023
6373
9310
17.2
600.
1648
3865
663
7117
7350
5293
8713
.760
0.15
6337
956
5934
1749
4753
6947
14.1
650.
2359
3228
476
1614
2382
3520
3610
.965
0.21
6932
023
6945
1427
5136
2000
11.3
700.
3206
2466
879
0810
3571
2096
548.
570
0.30
5925
078
7672
1062
0821
9249
8.7
750.
4380
1676
073
4065
449
1060
836.
375
0.41
7617
406
7268
6885
711
3041
6.5
80+
1.00
0094
2094
2040
634
4063
44.
380
+1.
0000
1013
710
137
4418
444
184
4.4
(Con
tinu
edon
next
page
)
April–June 2010, Volume 43, Number 2 69
TA
BL
E8.
New
Lif
eTa
bles
for
the
Whi
teP
opul
atio
nof
the
Uni
ted
Stat
es,1
780–
1900
(Con
tinue
d)
Whi
tem
ales
,181
0–19
Whi
tefe
mal
es,1
810–
19
Age
q xl x
d xL
xT
xe x
Age
q xl x
d xL
xT
xe x
00.
2190
1000
0021
898
8532
839
7500
739
.80
0.19
3810
0000
1938
487
400
4015
840
40.2
10.
0476
7810
237
2175
906
3889
679
49.8
10.
0460
8061
637
0778
428
3928
440
48.7
20.
0206
7438
015
3373
568
3813
773
51.3
20.
0221
7690
816
9776
009
3850
011
50.1
30.
0133
7284
796
572
345
3740
206
51.3
30.
0140
7521
110
5174
665
3774
002
50.2
40.
0104
7188
274
971
492
3667
860
51.0
40.
0112
7416
082
973
729
3699
338
49.9
50.
0302
7113
321
4735
0295
3596
368
50.6
50.
0311
7333
122
8336
0950
3625
608
49.4
100.
0232
6898
516
0134
0925
3246
073
47.1
100.
0242
7104
917
2235
0938
3264
659
45.9
150.
0365
6738
524
6133
0770
2905
148
43.1
150.
0418
6932
728
9733
9392
2913
720
42.0
200.
0492
6492
331
9331
6635
2574
378
39.7
200.
0572
6643
037
9832
2655
2574
329
38.8
250.
0505
6173
131
1930
0856
2257
743
36.6
250.
0606
6263
237
9530
3673
2251
674
36.0
300.
0503
5861
229
4628
5694
1956
887
33.4
300.
0605
5883
735
6228
5282
1948
001
33.1
350.
0546
5566
630
3727
0737
1671
192
30.0
350.
0604
5527
533
3926
8029
1662
719
30.1
400.
0616
5262
932
4225
5041
1400
455
26.6
400.
0658
5193
634
1725
1138
1394
690
26.9
450.
0736
4938
736
3323
7853
1145
414
23.2
450.
0729
4851
935
3623
3756
1143
552
23.6
500.
0900
4575
441
1821
8475
9075
6119
.850
0.09
1144
983
4096
2146
7490
9796
20.2
550.
1237
4163
651
5019
5306
6890
8516
.655
0.12
0940
887
4942
1920
7969
5122
17.0
600.
1697
3648
661
9016
6956
4937
8013
.560
0.16
0635
945
5772
1652
9550
3043
14.0
650.
2412
3029
673
0613
3214
3268
2410
.865
0.22
1530
173
6682
1341
5833
7749
11.2
700.
3254
2299
074
8196
246
1936
108.
470
0.31
0423
490
7292
9922
320
3590
8.7
750.
4416
1550
968
4960
420
9736
36.
375
0.42
1116
199
6822
6394
110
4367
6.4
80+
1.00
0086
5986
5936
943
3694
34.
380
+1.
0000
9377
9377
4042
640
426
4.3
Whi
tem
ales
,182
0–29
Whi
tefe
mal
es,1
820–
29
Age
q xl x
d xL
xT
xe x
Age
q xl x
d xL
xT
xe x
00.
2166
1000
0021
663
8548
639
9938
240
.00
0.19
1810
0000
1918
487
531
4038
406
40.4
10.
0472
7833
736
9476
157
3913
896
50.0
10.
0455
8081
636
8078
645
3950
875
48.9
20.
0204
7464
315
2373
836
3837
739
51.4
20.
0219
7713
716
8676
243
3872
230
50.2
30.
0131
7312
095
972
621
3763
903
51.5
30.
0138
7545
110
4574
908
3795
986
50.3
40.
0103
7216
074
571
773
3691
282
51.2
40.
0111
7440
682
473
978
3721
079
50.0
50.
0299
7141
621
3535
1740
3619
509
50.7
50.
0308
7358
322
6936
2239
3647
101
49.6
100.
0230
6928
015
9334
2420
3267
769
47.2
100.
0240
7131
317
1335
2283
3284
862
46.1
150.
0362
6768
824
5033
2314
2925
349
43.2
150.
0414
6960
028
8334
0793
2932
578
42.1
200.
0488
6523
831
8131
8236
2593
035
39.7
200.
0567
6671
737
8432
4125
2591
785
38.8
250.
0501
6205
731
1030
2509
2274
799
36.7
250.
0601
6293
337
8430
5206
2267
660
36.0
300.
0499
5894
729
4028
7384
1972
290
33.5
300.
0601
5914
935
5528
6858
1962
454
33.2
350.
0542
5600
730
3327
2451
1684
906
30.1
350.
0600
5559
433
3626
9629
1675
597
30.1
400.
0612
5297
432
4125
6766
1412
455
26.7
400.
0654
5225
834
1625
2748
1405
967
26.9
450.
0731
4973
336
3623
9576
1155
688
23.2
450.
0725
4884
135
3923
5359
1153
219
23.6
500.
0895
4609
741
2522
0175
9161
1319
.950
0.09
0645
302
4104
2162
5291
7860
20.3
550.
1231
4197
251
6519
6949
6959
3816
.655
0.12
0341
199
4956
1936
0370
1608
17.0
600.
1689
3680
762
1816
8490
4989
8913
.660
0.16
0036
242
5797
1667
1950
8005
14.0
650.
2404
3058
973
5313
4560
3304
9910
.865
0.22
0830
445
6722
1354
2034
1286
11.2
700.
3247
2323
575
4597
315
1959
388.
470
0.30
9823
723
7348
1002
4420
5866
8.7
750.
4411
1569
169
2161
150
9862
36.
375
0.42
0616
375
6887
6465
510
5622
6.5
80+
1.00
0087
6987
6937
473
3747
34.
380
+1.
0000
9487
9487
4096
640
966
4.3
(Con
tinu
edon
next
page
)
70 HISTORICAL METHODS
TA
BL
E8.
New
Lif
eTa
bles
for
the
Whi
teP
opul
atio
nof
the
Uni
ted
Stat
es,1
780–
1900
(Con
tinue
d)
Whi
tem
ales
,183
0–39
Whi
tefe
mal
es,1
830–
39
Age
q xl x
d xL
xT
xe x
Age
q xl x
d xL
xT
xe x
00.
2119
1000
0021
188
8580
440
4912
140
.50
0.18
7810
0000
1877
987
794
4084
478
40.8
10.
0462
7881
236
3976
665
3963
317
50.3
10.
0446
8122
136
2479
083
3996
684
49.2
20.
0200
7517
315
0274
377
3886
652
51.7
20.
0214
7759
716
6276
716
3917
601
50.5
30.
0129
7367
194
773
178
3812
275
51.7
30.
0136
7593
510
3175
399
3840
884
50.6
40.
0101
7272
473
672
341
3739
097
51.4
40.
0109
7490
481
374
481
3765
485
50.3
50.
0293
7198
821
1135
4664
3666
756
50.9
50.
0303
7409
122
4336
4849
3691
004
49.8
100.
0226
6987
815
7634
5448
3312
092
47.4
100.
0236
7184
916
9435
5007
3326
154
46.3
150.
0355
6830
224
2733
5441
2966
644
43.4
150.
0407
7015
428
5534
3633
2971
147
42.4
200.
0479
6587
531
5632
1484
2631
202
39.9
200.
0558
6729
937
5432
7110
2627
514
39.0
250.
0493
6271
930
9130
5867
2309
718
36.8
250.
0592
6354
537
6130
8322
2300
404
36.2
300.
0491
5962
829
2729
0822
2003
852
33.6
300.
0592
5978
435
4129
0066
1992
083
33.3
350.
0533
5670
130
2527
5942
1713
030
30.2
350.
0592
5624
333
2927
2892
1702
017
30.3
400.
0603
5367
632
3826
0285
1437
089
26.8
400.
0645
5291
434
1525
6033
1429
124
27.0
450.
0721
5043
836
3924
3093
1176
803
23.3
450.
0716
4949
935
4423
8636
1173
091
23.7
500.
0884
4679
941
3822
3651
9337
1020
.050
0.08
9645
955
4117
2194
8393
4455
20.3
550.
1218
4266
151
9520
0318
7100
5916
.655
0.11
9241
838
4985
1967
2671
4972
17.1
600.
1675
3746
662
7517
1642
5097
4213
.660
0.15
8736
853
5847
1696
4651
8245
14.1
650.
2388
3119
174
4913
7334
3380
9910
.865
0.21
9431
005
6803
1380
2034
8600
11.2
700.
3233
2374
276
7599
525
2007
658.
570
0.30
8424
203
7464
1023
5221
0580
8.7
750.
4400
1606
870
7062
664
1012
406.
375
0.41
9516
738
7022
6613
610
8227
6.5
80+
1.00
0089
9889
9838
577
3857
74.
380
+1.
0000
9716
9716
4209
242
092
4.3
Whi
tem
ales
,184
0–49
Whi
tefe
mal
es,1
840–
49
Age
q xl x
d xL
xT
xe x
Age
q xl x
d xL
xT
xe x
00.
2286
1000
0022
862
8468
238
7722
738
.80
0.20
2010
0000
2020
586
867
3925
388
39.3
10.
0496
7713
838
2974
879
3792
544
49.2
10.
0478
7979
538
1877
543
3838
521
48.1
20.
0215
7330
915
7372
475
3717
666
50.7
20.
0229
7597
817
4375
054
3760
978
49.5
30.
0138
7173
698
971
222
3645
190
50.8
30.
0145
7423
510
7873
674
3685
924
49.7
40.
0108
7074
776
770
348
3573
969
50.5
40.
0116
7315
784
972
716
3612
250
49.4
50.
0314
6998
021
9434
4415
3503
620
50.1
50.
0323
7230
823
3435
5705
3539
534
49.0
100.
0241
6778
616
3233
4849
3159
205
46.6
100.
0251
6997
417
5734
5476
3183
829
45.5
150.
0379
6615
425
0432
4509
2824
356
42.7
150.
0432
6821
729
4933
3710
2838
353
41.6
200.
0509
6365
032
3831
0154
2499
847
39.3
200.
0590
6526
838
5331
6705
2504
643
38.4
250.
0522
6041
231
5229
4179
2189
693
36.2
250.
0624
6141
438
3529
7485
2187
938
35.6
300.
0518
5726
029
6727
8882
1895
514
33.1
300.
0623
5758
035
8627
8935
1890
452
32.8
350.
0562
5429
330
4926
3842
1616
632
29.8
350.
0620
5399
433
5026
1597
1611
517
29.8
400.
0633
5124
432
4324
8114
1352
790
26.4
400.
0674
5064
534
1524
4685
1349
920
26.7
450.
0754
4800
136
2123
0955
1104
676
23.0
450.
0746
4722
935
2222
7341
1105
236
23.4
500.
0921
4438
140
8621
1688
8737
2119
.750
0.09
3043
707
4064
2083
7687
7894
20.1
550.
1262
4029
450
8418
8762
6620
3316
.455
0.12
3139
643
4880
1860
1766
9518
16.9
600.
1725
3521
060
7316
0870
4732
7113
.460
0.16
3134
764
5669
1596
4648
3500
13.9
650.
2442
2913
771
1512
7899
3124
0210
.765
0.22
4129
095
6521
1291
7332
3854
11.1
700.
3282
2202
272
2792
045
1845
038.
470
0.31
3022
574
7065
9520
919
4681
8.6
750.
4437
1479
665
6557
565
9245
86.
275
0.42
3115
509
6562
6114
199
472
6.4
80+
1.00
0082
3182
3134
893
3489
34.
280
+1.
0000
8947
8947
3833
138
331
4.3
(Con
tinu
edon
next
page
)
April–June 2010, Volume 43, Number 2 71
TA
BL
E8.
New
Lif
eTa
bles
for
the
Whi
teP
opul
atio
nof
the
Uni
ted
Stat
es,1
780–
1900
(Con
tinue
d)
Whi
tem
ales
,185
0–59
Whi
tefe
mal
es,1
850–
59
Age
q xl x
d xL
xT
xe x
Age
q xl x
d xL
xT
xe x
00.
2465
1000
0024
647
8348
636
8494
136
.80
0.22
2010
0000
2219
885
572
3713
004
37.1
10.
0542
7535
340
8772
941
3601
455
47.8
10.
0532
7780
241
4175
359
3627
432
46.6
20.
0236
7126
516
7970
376
3528
514
49.5
20.
0254
7366
118
7172
669
3552
073
48.2
30.
0151
6958
710
5269
039
3458
138
49.7
30.
0161
7179
011
5871
188
3479
403
48.5
40.
0118
6853
481
068
113
3389
099
49.5
40.
0128
7063
290
570
162
3408
215
48.3
50.
0339
6772
422
9633
2882
3320
986
49.0
50.
0353
6972
824
6434
2478
3338
053
47.9
100.
0254
6542
916
6032
2992
2988
104
45.7
100.
0268
6726
318
0133
1815
2995
575
44.5
150.
0397
6376
825
3431
2505
2665
112
41.8
150.
0457
6546
329
9131
9835
2663
760
40.7
200.
0538
6123
432
9529
7932
2352
607
38.4
200.
0623
6247
138
9230
2628
2343
925
37.5
250.
0555
5793
932
1328
1661
2054
676
35.5
250.
0661
5858
038
7028
3224
2041
298
34.8
300.
0557
5472
630
5026
6003
1773
015
32.4
300.
0662
5471
036
2326
4493
1758
074
32.1
350.
0605
5167
631
2725
0561
1507
012
29.2
350.
0661
5108
733
7924
6989
1493
581
29.2
400.
0679
4854
932
9523
4505
1256
451
25.9
400.
0716
4770
834
1722
9998
1246
592
26.1
450.
0805
4525
336
4121
7165
1021
946
22.6
450.
0792
4429
135
1021
2680
1016
594
23.0
500.
0977
4161
340
6619
7899
8047
8119
.350
0.09
8440
781
4012
1938
7580
3914
19.7
550.
1326
3754
749
8017
5285
6068
8216
.255
0.12
9536
769
4762
1719
3761
0040
16.6
600.
1790
3256
758
2814
8264
4315
9713
.360
0.16
9832
006
5434
1464
4643
8102
13.7
650.
2502
2673
966
9011
6968
2833
3310
.665
0.23
1326
572
6145
1174
9929
1656
11.0
700.
3331
2004
966
7883
549
1663
658.
370
0.31
9020
427
6516
8584
617
4157
8.5
750.
4476
1337
159
8551
893
8281
66.
275
0.42
7613
911
5948
5468
688
311
6.3
80+
1.00
0073
8673
8630
923
3092
34.
280
+1.
0000
7963
7963
3362
533
625
4.2
Whi
tem
ales
,186
0–69
Whi
tefe
mal
es,1
860–
69
Age
q xl x
d xL
xT
xe x
Age
q xl x
d xL
xT
xe x
00.
2071
1000
0020
708
8612
535
0691
735
.10
0.18
9310
0000
1893
087
696
4057
822
40.6
10.
0469
7929
237
1877
098
3420
791
43.1
10.
0467
8107
037
8378
838
3970
126
49.0
20.
0206
7557
315
5674
749
3343
693
44.2
20.
0223
7728
717
2376
374
3891
288
50.3
30.
0132
7401
798
073
508
3268
945
44.2
30.
0142
7556
510
7675
005
3814
913
50.5
40.
0103
7303
775
372
645
3195
437
43.8
40.
0113
7448
984
074
052
3739
908
50.2
50.
0295
7228
421
3335
6085
3122
792
43.2
50.
0311
7364
922
8836
2524
3665
856
49.8
100.
0216
7015
015
1334
6968
2766
706
39.4
100.
0230
7136
116
3835
2709
3303
332
46.3
150.
0812
6863
755
7232
9256
2419
738
35.3
150.
0392
6972
327
3134
1787
2950
623
42.3
200.
1049
6306
566
1329
8795
2090
482
33.1
200.
0540
6699
236
1632
5919
2608
837
38.9
250.
0999
5645
256
3926
8165
1791
687
31.7
250.
0582
6337
636
8730
7660
2282
917
36.0
300.
1041
5081
452
9124
0840
1523
522
30.0
300.
0594
5968
835
4528
9580
1975
257
33.1
350.
0927
4552
242
2221
7057
1282
682
28.2
350.
0602
5614
333
7827
2272
1685
677
30.0
400.
0831
4130
134
3219
7924
1065
625
25.8
400.
0657
5276
534
6525
5163
1413
406
26.8
450.
0819
3786
931
0318
1586
8677
0122
.945
0.07
3549
300
3626
2374
3611
5824
223
.550
0.09
4334
766
3279
1656
3068
6115
19.7
500.
0922
4567
442
0921
7849
9208
0620
.255
0.12
5131
487
3938
1475
8952
0484
16.5
550.
1225
4146
550
8119
4623
7029
5717
.060
0.16
9727
549
4675
1260
5737
2896
13.5
600.
1617
3638
458
8516
7208
5083
3414
.065
0.23
9422
874
5476
1006
7824
6839
10.8
650.
2228
3049
967
9613
5506
3411
2611
.270
0.32
2917
397
5617
7294
314
6161
8.4
700.
3101
2370
373
5110
0139
2056
208.
775
0.44
0211
780
5186
4593
673
218
6.2
750.
4205
1635
268
7664
570
1054
816.
580
+1.
0000
6594
6594
2728
227
282
4.1
80+
1.00
0094
7694
7640
910
4091
04.
3(C
onti
nued
onne
xtpa
ge)
72 HISTORICAL METHODS
TA
BL
E8.
New
Lif
eTa
bles
for
the
Whi
teP
opul
atio
nof
the
Uni
ted
Stat
es,1
780–
1900
(Con
tinue
d)
Whi
tem
ales
,187
0–79
Whi
tefe
mal
es,1
870–
79
Age
q xl x
d xL
xT
xe x
Age
q xl x
d xL
xT
xe x
00.
1753
1000
0017
533
8825
343
9889
944
.00
0.16
1610
0000
1616
489
493
4387
395
43.9
10.
0406
8246
733
4880
491
4310
646
52.3
10.
0407
8383
634
1581
821
4297
902
51.3
20.
0180
7911
914
2278
365
4230
154
53.5
20.
0195
8042
115
6679
591
4216
081
52.4
30.
0116
7769
690
077
228
4151
790
53.4
30.
0125
7885
598
778
341
4136
491
52.5
40.
0090
7679
669
176
436
4074
562
53.1
40.
0099
7786
876
977
468
4058
149
52.1
50.
0257
7610
519
5837
5629
3998
125
52.5
50.
0272
7709
920
9938
0246
3980
681
51.6
100.
0185
7414
713
6936
7311
3622
496
48.9
100.
0197
7500
014
8037
1300
3600
435
48.0
150.
0291
7277
821
2035
8587
3255
185
44.7
150.
0337
7352
024
7536
1412
3229
135
43.9
200.
0409
7065
728
8834
6066
2896
598
41.0
200.
0468
7104
533
2734
6906
2867
723
40.4
250.
0434
6776
929
4133
1493
2550
532
37.6
250.
0512
6771
734
6532
9924
2520
817
37.2
300.
0455
6482
829
4931
6767
2219
039
34.2
300.
0530
6425
234
0831
2742
2190
893
34.1
350.
0506
6187
931
2930
1571
1902
272
30.7
350.
0544
6084
433
1029
5946
1878
151
30.9
400.
0574
5875
033
7328
5316
1600
701
27.2
400.
0598
5753
434
4127
9069
1582
205
27.5
450.
0689
5537
738
1626
7343
1315
385
23.8
450.
0677
5409
436
6326
1310
1303
136
24.1
500.
0851
5156
043
8724
6834
1048
042
20.3
500.
0856
5043
143
1724
1360
1041
825
20.7
550.
1172
4717
355
2922
2044
8012
0817
.055
0.11
5046
113
5304
2173
0780
0466
17.4
600.
1602
4164
566
7219
1542
5791
6413
.960
0.15
3140
810
6247
1884
3158
3158
14.3
650.
2284
3497
279
8915
4888
3876
2211
.165
0.21
3534
563
7379
1543
6539
4728
11.4
700.
3124
2698
384
2911
3844
2327
358.
670
0.30
0527
183
8167
1154
9924
0362
8.8
750.
4323
1855
480
2172
720
1188
916.
475
0.41
2719
016
7847
7546
212
4863
6.6
80+
1.00
0010
533
1053
346
171
4617
14.
480
+1.
0000
1116
911
169
4940
149
401
4.4
Whi
tem
ales
,188
0–89
Whi
tefe
mal
es,1
880–
89
Age
q xl x
d xL
xT
xe x
Age
q xl x
d xL
xT
xe x
00.
1494
1000
0014
936
8999
347
0822
447
.10
0.13
1710
0000
1316
891
441
4795
214
48.0
10.
0354
8506
430
0883
290
4618
231
54.3
10.
0340
8683
229
5085
091
4703
773
54.2
20.
0158
8205
612
9681
370
4534
941
55.3
20.
0163
8388
213
6583
158
4618
682
55.1
30.
0102
8076
182
380
333
4453
571
55.1
30.
0105
8251
786
882
065
4535
524
55.0
40.
0079
7993
863
179
609
4373
238
54.7
40.
0083
8164
967
681
297
4453
458
54.5
50.
0225
7930
717
8639
2067
4293
629
54.1
50.
0228
8097
318
4840
0245
4372
161
54.0
100.
0158
7752
012
2838
4531
3901
562
50.3
100.
0162
7912
512
8239
2420
3971
916
50.2
150.
0251
7629
219
1137
6682
3517
031
46.1
150.
0276
7784
321
5238
3834
3579
496
46.0
200.
0357
7438
126
5736
5261
3140
349
42.2
200.
0389
7569
129
4137
1101
3195
662
42.2
250.
0384
7172
427
5635
1729
2775
089
38.7
250.
0431
7275
031
3635
5908
2824
561
38.8
300.
0411
6896
828
3133
7761
2423
360
35.1
300.
0455
6961
431
6534
0155
2468
652
35.5
350.
0461
6613
730
5032
3058
2085
599
31.5
350.
0473
6644
931
4332
4385
2128
497
32.0
400.
0526
6308
733
2130
7129
1762
541
27.9
400.
0525
6330
633
2030
8227
1804
112
28.5
450.
0636
5976
538
0028
9325
1455
412
24.4
450.
0602
5998
536
1029
0900
1495
886
24.9
500.
0791
5596
544
2926
8753
1166
086
20.8
500.
0770
5637
543
3827
1030
1204
985
21.4
550.
1098
5153
656
5824
3536
8973
3317
.455
0.10
4852
037
5452
2465
5393
3955
17.9
600.
1511
4587
869
3321
2057
6537
9714
.360
0.14
1146
584
6575
2164
8468
7402
14.8
650.
2176
3894
584
7617
3534
4417
4011
.365
0.20
0340
009
8013
1800
1347
0917
11.8
700.
3017
3046
991
9312
9361
2682
068.
870
0.28
6631
996
9169
1370
5729
0904
9.1
750.
4241
2127
690
2383
821
1388
456.
575
0.40
1122
827
9156
9124
315
3848
6.7
80+
1.00
0012
253
1225
355
024
5502
44.
580
+1.
0000
1367
113
671
6260
462
604
4.6
(Con
tinu
edon
next
page
)
April–June 2010, Volume 43, Number 2 73
TA
BL
E8.
New
Lif
eTa
bles
for
the
Whi
teP
opul
atio
nof
the
Uni
ted
Stat
es,1
780–
1900
(Con
tinue
d)
Whi
tem
ales
,189
0–99
Whi
tefe
mal
es,1
890–
99
Age
q xl x
d xL
xT
xe x
Age
q xl x
d xL
xT
xe x
00.
1305
1000
0013
052
9125
549
4801
349
.50
0.11
0610
0000
1105
992
812
5121
250
51.2
10.
0316
8694
827
5085
326
4856
758
55.9
10.
0292
8894
125
9987
407
5028
438
56.5
20.
0142
8419
811
9983
563
4771
433
56.7
20.
0140
8634
212
0985
701
4941
031
57.2
30.
0092
8299
976
482
602
4687
870
56.5
30.
0091
8513
277
584
730
4855
330
57.0
40.
0071
8223
558
481
931
4605
268
56.0
40.
0071
8435
860
384
044
4770
601
56.6
50.
0202
8165
116
4940
4130
4523
337
55.4
50.
0196
8375
516
4641
4661
4686
556
56.0
100.
0139
8000
111
0939
7235
4119
207
51.5
100.
0136
8211
011
1740
7756
4271
895
52.0
150.
0219
7889
217
3139
0135
3721
972
47.2
150.
0231
8099
318
7340
0281
3864
139
47.7
200.
0318
7716
224
5437
9672
3331
837
43.2
200.
0328
7912
025
9438
9113
3463
858
43.8
250.
0347
7470
725
9036
7061
2952
165
39.5
250.
0369
7652
628
2537
5565
3074
744
40.2
300.
0378
7211
727
2535
3773
2585
104
35.8
300.
0396
7370
129
2036
1203
2699
179
36.6
350.
0429
6939
229
7533
9522
2231
331
32.2
350.
0418
7078
129
5734
6511
2337
976
33.0
400.
0492
6641
732
6632
3918
1891
809
28.5
400.
0467
6782
431
6533
1204
1991
465
29.4
450.
0597
6315
137
6830
6333
1567
890
24.8
450.
0542
6465
835
0531
4527
1660
260
25.7
500.
0748
5938
244
3928
5814
1261
557
21.2
500.
0700
6115
342
8129
5060
1345
733
22.0
550.
1043
5494
357
2926
0394
9757
4417
.855
0.09
6456
871
5485
2706
4610
5067
318
.560
0.14
4149
214
7092
2283
4271
5350
14.5
600.
1311
5138
767
3924
0086
7800
2715
.265
0.20
8942
123
8801
1886
1048
7008
11.6
650.
1889
4464
884
3520
2151
5399
4112
.170
0.29
2833
321
9758
1422
1329
8398
9.0
700.
2741
3621
399
2615
6250
3377
909.
375
0.41
7223
564
9831
9324
215
6185
6.6
750.
3903
2628
710
260
1057
8518
1540
6.9
80+
1.00
0013
733
1373
362
943
6294
34.
680
+1.
0000
1602
716
027
7575
475
754
4.7
74 HISTORICAL METHODS
TABLE 9. Proportion Dying in Age Interval x to x + n, White Males
Exact age, x n 1860 1861 1862 1863 1864 1865 1866–69
0 1 0.2071 0.2071 0.2071 0.2071 0.2071 0.2071 0.20711 1 0.0469 0.0469 0.0469 0.0469 0.0469 0.0469 0.04692 1 0.0206 0.0206 0.0206 0.0206 0.0206 0.0206 0.02063 1 0.0132 0.0132 0.0132 0.0132 0.0132 0.0132 0.01324 1 0.0103 0.0103 0.0103 0.0103 0.0103 0.0103 0.01035 5 0.0295 0.0295 0.0295 0.0295 0.0295 0.0295 0.0295
10 5 0.0216 0.0216 0.0216 0.0216 0.0216 0.0216 0.021615 5 0.0339 0.0434 0.1300 0.1507 0.1682 0.1451 0.033920 5 0.0468 0.0549 0.1398 0.1821 0.2518 0.1703 0.046825 5 0.0491 0.0587 0.1440 0.1713 0.2230 0.1445 0.049130 5 0.0505 0.0588 0.1383 0.1755 0.2391 0.1617 0.050535 5 0.0555 0.0605 0.1118 0.1403 0.1927 0.1411 0.055540 5 0.0626 0.0656 0.0900 0.1069 0.1410 0.1148 0.062645 5 0.0747 0.0762 0.0829 0.0892 0.1032 0.0957 0.074750 5 0.0915 0.0955 0.0962 0.0971 0.0992 0.0990 0.091555 5 0.1251 0.1251 0.1251 0.1251 0.1251 0.1251 0.125160 5 0.1697 0.1697 0.1697 0.1697 0.1697 0.1697 0.169765 5 0.2394 0.2394 0.2394 0.2394 0.2394 0.2394 0.239470 5 0.3229 0.3229 0.3229 0.3229 0.3229 0.3229 0.322975 5 0.4402 0.4402 0.4402 0.4402 0.4402 0.4402 0.440280+ 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
from table 8 with a closely corresponding Haines life table.Figure 7 compares the 1870–79 life table for white males(e0 = 44.0) with Haines’s 1870 life table (e0 = 44.1). Ingeneral, there is close correspondence between the two age
profiles. The 1870–79 life table indicates higher mortalityrates between ages 10 and 30, but the difference is modest:age-specific mortality rates are 12 percent higher at age 20than in Haines’s table. Figure 8 compares the 1840–49 life
34.0
36.0
38.0
40.0
42.0
44.0
46.0
48.0
50.0
52.0
54.0
17901800
18101820
18301840
18501860
18701880
18901900
Year
Life
exp
ecta
ncy
at b
irth
(ye
ars)
New es�mates, white males
New es�mates, white females
Haines U.S. model, white males
Haines U.S. model, white females
FIGURE 6. Life expectancy at birth, white population of the United States, 1790–1900.
April–June 2010, Volume 43, Number 2 75
100
1,000
10,000
Age group
Pro
port
ion
dyin
g, 1
0,00
0 q x
(lo
g sc
ale)
Haines 1870 white males,U.S. model
New estimates, white males,1870-79
0 1 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
FIGURE 7. Proportion dying by age group, white males.
table for white females (e0 = 40.6) with Haines’s 1850 lifetable (e0 = 40.6). Again, with the exception of the part ofthe profile between adolescence and middle age, there isclose correspondence between the two age profiles. The dif-ference between the curves between ages 10 and 35, however,
is much greater. At age 20 the 1840–49 life table suggestsa mortality rate 27 percent higher than Haines’s 1850 lifetable.
Figure 8 also includes a modified plot of the proportionsdying in Haines’s 1850 female life table. The age-specific
100
1,000
10,000
Age group
Pro
port
ion
dyin
g, 1
0,00
0 q x
(lo
g sc
ale)
Haines 1850, white females,U.S. model
New estimates, white females,1840-49
Haines 1850, white females,adjusted
0 1 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
FIGURE 8. Proportion dying by age group, white females. qx = proportion dying in age interval.
76 HISTORICAL METHODS
death rates in the Haines table were modified by doublingcause-specific death rates attributed to pulmonary tubercu-losis and maternal mortality reported by Samuel Preston fornational populations with life expectancy at birth less than 45years (1976). The adjusted profile corresponds very closelywith the age profile of female mortality in the 1840–49 lifetable. Although speculative and seemingly large, the ad-justments correspond with what we know about changesin mortality and fertility between the mid-nineteenth andearly twentieth centuries. As discussed earlier, mortality fromtuberculosis fell rapidly in the late nineteenth century. Fer-tility among women in the 1900–1902 DRA life table wasapproximately half that of women in the 1840s. Although wecannot know the true age-specific mortality rates for whitewomen in the 1840s, it is likely that the profile differed fromthat in the 1900–1902 DRA in the way indicated.
All of the decennial tables in table 8 are based, of course,on assumptions with substantial risk of error. Much moreresearch is needed on biases in demographic estimation fromgenealogical sources. Although based in part on comparisonwith other sources and in part on the suspected impact ofmigration censoring and selection biases, the crude assump-tions about the overestimation of adult life expectancy ingenealogical-based estimates and sex differentials in adultlife expectancy made in tables 4 and 5 are sources of po-tential error. Another weakness is the required method ofinferring a complete life table from a single parameter, lifeexpectancy at age 20. Historical demographers in Europeand elsewhere have called attention to the changing relation-ship between infant, childhood, and adult mortality over thecourse of the nineteenth century (Woods 1993). It is unlikelythat the United States was an exception. Empirical researchon infant and childhood mortality in the United States issorely needed. Source material, however, remains a majorissue.
Despite these caveats, the life tables shown in table 8should prove useful for a wide variety of historical research.In addition to capturing known mortality trends not reflectedin existing life tables, they more accurately represent thelikely sex- and age-specific profile of nineteenth-centurymortality. The life tables should also prove useful as a point ofreference for subsequent studies and critiques. With any luck,nineteenth-century demographers will have more choices oflife tables with a firmer empirical base in the not too distantfuture.
NOTES
1. This work was supported in part by NIHCD grant number 1 K01-HD052617–01 and an Arthur H. Cole Grant-in-Aid Award from the Eco-nomic History Association. The author would like to thank Samuel H.Preston, Douglas Ewbank, and Michael R. Haines for helpful comments.
2. The original Death Registration Area (DRA) included only 10 statesand the District of Columbia. The system was deemed complete in 1933,when Texas was added to the system, although considerable underreportingof births and deaths continued to plague the system until the 1940s.
3. Although Massachusetts’s death registration system was implementedin 1842, it took several years for the system to become effective. By 1860,
Robert Gutman has estimated that only 8 percent of deaths were unrecorded(Vinovskis 1972, 186).
4. Condran and Crimmins’s (1980, 188–90) application of the ChandraSekar-Deming technique suggests that approximately 85 percent of deathsin rural areas and 92 percent of deaths in urban areas were registered. Infantdeaths were missed more often than deaths at other ages.
5. Other potential problems include the possibility that the deaths ofchildren ages 5–19, while more fully enumerated than deaths at other ages,were still underreported, and the possibility of a changing level of undercountfrom census to census. If underreporting was significant, the Haines (1998)life tables may overstate life expectancy. The addition of some state deathregisters in 1880 likely lowered the overall undercount and may explainsome of the sharp decline in life expectancy between the 1870 and 1880estimates.
6. Kasakoff and Adams (1995) report the average age at death by birthcohort, not period. In the figure, the cohort estimates are offset 20 years toincrease comparability.
7. The urban population is defined liberally as all individuals livingin urbanized areas and in all places of 2,500 or more residents outsideof urbanized areas. The percentage living in large cities with significantsanitation problems was much smaller. The urban population increasedfrom 6.1 percent in 1800 to 10.8 percent in 1840, 28.2 percent in 1880, and51.2 percent in 1920. By the turn of the century, when urbanization wassignificant enough to pose a major impact on national life expectancy, thepublic health movement had made significant strides in introducing cleanwater supplies, sewer systems, and other public health projects, greatlyreducing the urban-rural differential in life expectancy.
8. Considerable uncertainties surround estimates of real national incomein the early nineteenth century. Most economic historians conclude that therewas a sharp increase in real economic growth in the 1820s. According toRichard Sutch, the annual growth rate between 1800 and 1828 averagedabout 0.6 percent per year. Between 1828 and 1860 it averaged more thantwice that rate (Sutch 2006).
9. Under some conditions, censoring bias does not impart a downwardbias in life expectancy estimates. If a researcher knows when an individualdisappeared from observation and if censored individuals experienced thesame risk of death as noncensored individuals, for example, it is possible toconstruct nonbiased age-specific mortality estimates. Relative to the exten-sive rules followed by analysts of community-based reconstitution studies,however, researchers relying on genealogical data have shown little inter-est in precisely determining when the population was under observation.Neither Kunze (1979) nor Pope (1992) appeared to have included risk yearsfrom right-censored individuals in the calculation of age-specific death rates.Only individuals with known birth and death dates are included. Given theseselection criteria, censoring bias will impart a downward bias (for an ex-tended rumination on biases in early American mortality studies, see Smith1979).
10. The average of Haines’s (1998) 1850 and 1860 U.S. model census-based estimates of life expectancy at age 20 was assumed to be representativeof the 1850s, the 1870 and 1880 estimates representative of the 1870s, andthe 1880 and 1890 estimates representative of the 1880s. It was not assumedthat the average of the 1860 and 1870 estimates would be representative ofthe 1860s, however, because the census-based estimates fail to consider theimpact of the U.S. Civil War (1861–65).
11. Although genealogies are successful in tracking some family mem-bers across time and space, migrating family members are more prone tobe lost from observation. Hall and Ruggles (2004) have shown that internalmigration in the United States exhibited a “U-shaped” pattern between 1850and 2000. Almost one-in-two whites age 50–59 between 1850 and 1880were living in a state other than their birth state. This ratio dropped steadilyafter 1880, reached a low of about one-in-three in the period 1940–70 andthen increased to over four-in-ten in the 2000 census.
12. Although Kunze’s (1979, 200) sample appears to be slightly largerthan Pope’s (1992, 282) sample, Kunze does not report the number of casesused in his period estimates. The combined estimates shown in table 5 aretherefore unweighted averages, smoothed slightly in the period before 1850.
13. Estimates of the white birth rate were obtained with stable populationmethods, the published age distributions of the 1800 census, and life tablesconstructed by fitting the adjusted and unadjusted Pope (1992) and Kunze(1979) estimate of life expectancy at age 20 to the 1901 rural DRA life tableas described in the latter part of this article.
14. Males and females enter Pope’s (1992) sample as either a childof bloodline parents or as a spouse of a bloodline individual. The former
April–June 2010, Volume 43, Number 2 77
contribute risk years from birth to death while the latter contribute risk yearsfrom marriage to death. Theoretically, there should be approximately equalnumbers of men and women in the samples. According to Pope’s illustrationof a “typical” family history, however, 11 percent of men in the genealogicalsamples had a missing birth date and 43 percent a missing death date. Thepercentages for women were 16 and 59 percent, respectively (ibid., 273). Asa result, Pope’s period life expectancy estimates are based on 3,166 malesand 2,338 females with known birth and death dates (ibid., 282). Kunze(1979, 200–204) does not discuss the completeness of his demographic databy sex, but similar differences are apparent in the number of males andfemales used in his analysis. Kasakoff and Adams (1995) report only themean age at death of males.
15. Stolnitz’s (1956, 23–25) classic review of long-term mortality trendscalled explicit attention to instances of higher female mortality in the pre-vious century. Although females in Western countries between 1840 and1910 typically enjoyed lower mortality rates during infancy and older ages,higher female mortality rates from late childhood through most of the child-bearing years was common. The modern pattern of lower female mortalityat all ages did not become typical until the 1930s. Although the life ta-bles Stolnitz examined tended to favor higher female life expectancy at allages, higher male life expectancy could be found across an “appreciable”range of ages well into the twentieth century in Ireland, Italy, Austria, andBulgaria.
16. Stolnitz (1956, 23–25) reported the largest persisting female disad-vantages in life expectancy among the Irish population, which experiencedhigh fertility, low nutritional status, preferential treatment for males, andendemic tuberculosis well into the twentieth century (ibid.; Kennedy 1973).
17. The results also suggest that Haines’s (1998) life tables overstatefemale life expectancy at age 20 relative to male life expectancy. The relativeoverstatement is likely a result of Haines’s choice of model life tables—a“U.S. model” constructed from the 1900–1902 DRA and Coale, Demeny,and Vaughan’s (1966) west model. Both models are based on the mortalityexperience of more urban and lower fertility populations than the nineteenth-century population of the United States. As discussed at greater length inthe section on the age profile of nineteenth-century mortality, these modelslikely understate female mortality during childbearing years relative to otherages and overstate female life expectancy at age 20 relative to male lifeexpectancy.
18. Details on weighting and combining the 1900–1902 overall and ruralDRA life tables can be found in the section on new decennial life tables andin note 24 below.
19. Four- and five-parameter models have also been proposed (see, e.g.,Ewbank, de Leon, and Stoto 1983).
20. Although Coale and Zelnik (1963, 168–69) observed a good corre-spondence between the 138 life tables that were used to construct the westmodel and the 1900–1902 DRA life table, only 36 of the 138 life tables camefrom nineteenth-century populations. The model matches the male experi-ence better than the female experience. Coale and Zelnik did not comparethe 1900–1902 rural DRA life table with the model.
21. Rebecca Kippen (2005) has noted that maternal deaths are often un-derreported in official statistics and in estimates derived from family recon-stitution studies. Her revised estimates of maternal mortality for nineteenth-century Tasmania—7 deaths per 1,000 live births—are approximately twiceas high as estimates derived from other sources. Even so, maternal mortalityremained a distant second leading cause of death among women age 29–44behind pulmonary tuberculosis.
22. It is important to remember that the 1900–1902 overall DRA in-cluded females in the rural DRA. The differences would have been greaterif we were able to compare urban females directly to rural females (for ananalysis of urban-rural mortality differentials in 1890 and 1900 see Condranand Crimmins 1980).
23. Typically, sex mortality differentials favor females at lower mortalitylevels. Sex differences in mortality between historical and modern popula-tions are the result of changes in causes of death associated with mortalitydecline. Female advantages in mortality at all ages emerged only with the de-cline of tuberculosis and other infectious diseases as leading causes of deathand their replacement with degenerative diseases. The decline of maternalmortality also played a small role (Preston 1976).
24. According to table 2, the 1900–1902 DRA life table was 60.1 per-cent urban and the 1900–1902 rural DRA life table was 13.2 percent ur-ban. If W1 is the weight needed for the overall DRA life table, W2 is theweight for the rural DRA life table, and the desired combined life tableis 40.2 percent urban, then (W1 × 60.1) + (W2 × 13.2) = 40.2. Further,
W1 + W2 = 1. Solving the second equation for W2, we get W2 = 1—W1.By substitution, the first equation becomes (W1 × 60.1) + ([1– W1] × 13.2)= 40.2. Solving for W1, we get 0.575. Substituting the result in the secondequation and solving yields 0.425 for W2.
25. Haines and Preston (1997, 77) state that the “improvement was mostrapid in large urban areas, where mortality had been the worst. The substan-tial urban mortality penalty . . . of the late nineteenth century was rapidlydisappearing by the early twentieth century. Public health improvements,better nutrition and shelter, and some advances in medical science all playeda role.”
26. Cohort differences between the male-female differential in 10-yearsurvivorship ratios in the 1860s relative to the average male-female differ-ential in 10-year survivorship ratios in the 1850s and 1870s were assumedto be because of the excess male mortality in the war. The estimate requiredfour major assumptions: (1) the native-born white population was closed tomigration; (2) changes in net census underenumeration had an equal impacton native-born white males and females; (3) foreign-born white men suf-fered rates of mortality in the war equal to the native-born white population;and (4) there were negligible civilian deaths among native-white women age15–45. For the approximately equal rates of mortality among foreign-bornand native-born men, see Lee (2003, 60). For the limited number of civil-ian casualties in the U.S. Civil War, see McPherson (1988, 619) and Neely(2007). Although the resulting estimate of approximately 713,000, excessmale deaths is larger than the 588,000 usually attributed to white men in thewar, there are many reasons to assume the 588,000 figure is too low (Hacker1999, chap. 2; Faust 2006).
27. The Union Army data set, collected by the University of ChicagoCenter for Population Economics and Brigham Young University under thedirection of Robert W. Fogel, was used to parse deaths by year and withinfive-year age groups.
REFERENCES
Abbott, S. W. 1899. A Massachusetts life table for the five years 1893–97. InThirteenth annual report of the State Board of Health of Massachusetts,797–827. Boston: Wright and Potter.
Alter, G., M. Manfredini, and P. Nystedt. 2004. Gender differences in mor-tality. In Life under pressure: Mortality and living standards in Europeand Asia, 1700–1900, ed., T. Bengtsson, C. Campbell, and J. Z. Lee,327–57. Cambridge, MA: The MIT Press.
Anderson, M. J. 1988. The American census: A social history. New Haven,CT: Yale University Press.
Anderton, D. L., and S. H. Leonard. 2004. Grammars of death: An analysisof nineteenth-century literal causes of death from the age of miasmas togerm theory. Social Science History 28:111–43.
Beemer, J. K., D. L. Anderton, and S. Hautaniemi Leonard. 2005. Sewers inthe city: A case study of individual-level mortality and public health ini-tiatives in Northampton, Massachusetts at the turn of the century. Journalof the History of Medicine and the Allied Sciences 60:42–72.
Billings, J. S. 1885. Report on the mortality and vital statistics of the UnitedStates as returned by the tenth census. U.S. Census of Population: 1880,vol. 10, part 1. Washington, DC: Government Printing Office.
———. 1886. U.S. report on the mortality and vital statistics of the UnitedStates as returned by the tenth census. U.S. Census of Population: 1880,vol. 10, part 2. Washington, DC: Government Printing Office.
Brass, W. 1971. On the scale of mortality. In Biological aspects of demog-raphy, ed. W. Brass. London: Taylor & Francis.
Campbell, C., J. Z. Lee, and T. Bengtsson. 2004. Economic stress andmortality. In Life under pressure: Mortality and living standards in Europeand Asia. ed., T. Bengtsson, C. Campbell, J. Z. Lee, 61–84.
Cheney, R. A. 1984. Seasonal aspects of infant and childhood mortal-ity: Philadelphia, 1865–1920. Journal of Interdisciplinary History 14:561–85.
Coale, A. J., P. Demeny, and B. Vaughan. 1966. Regional model life tablesand stable populations. Princeton, NJ: Princeton University Press.
Coale, A. J., and M. Zelnik. 1963. New estimates of fertility and populationin the United States: A study of annual white births from 1855 to 1960and of completeness of enumeration in the censuses from 1880 to 1960.Princeton, NJ: Princeton University Press.
78 HISTORICAL METHODS
Condran, G. A. 1987. Declining mortality in the United States in thelate nineteenth and early twentieth centuries. Annales de DemographieHistorique:119–37.
Condran, G. A., and R. A. Cheney. 1982. Mortality trends in Philadel-phia: Age- and cause-specific death rates 1870–1930. Demography 19:97–123.
Condran, G. A., and E. Crimmins. 1979. A description and evaluation of mor-tality data in the federal census: 1850–1900. Historical Methods 12:1–23.
———. 1980. Mortality differentials between rural and urban areas of statesin the northeastern United States, 1890–1900. Journal of Historical Ge-ography 6:179–202.
Costa, D. L., and R. H. Steckel. 1997. Long-term trends in health, welfare,and economic growth in the United States. In Health and Welfare dur-ing Industrialization, ed. R. H. Steckel and R. Floud, 47–89. Chicago:University of Chicago Press.
Crimmins, E. M. 1980. The completeness of 1900 mortality data collectedby registration and enumeration for rural and urban parts of states: Es-timates using the Chandra Sekar-Deming technique. Historical Methods13:163–69.
Davin, E. L. 1993. The era of the common child: Egalitarian death in ante-bellum America. Mid-America 75:134–63.
De Bow, J. D. B. 1855. Mortality statistics of the seventh census of theUnited States, 1850. Washington, DC: United States Census Office, A.O. P. Nicholson.
Duffy, J. 1990. The sanitarians: A history of American public health. Ur-bana: University of Illinois Press.
Easterlin, R. A. 1977. Population issues in American economic history:A survey and critique. In Recent developments in the study of businessand economic history: Essays in honor of Herman E. Krooss, ed. R. E.Gallman, 131–58. Greenwich, CT: JAI Press.
Elliott, E. B. 1858. On the law of human mortality that appears to obtainin Massachusetts, with tables of practical value deduced therefrom. InProceedings of the American Association for the Advancement of Science,ed. J. Lovering. 51–81. Cambridge, MA: Allen and Farnham.
———. 1874. An approximate life-table for the United States on the basisof the ninth census, 1870. In U.S. Census Office, Statistical atlas ofthe United States based on the results of the Ninth Census 1870, withcontributions from many eminent men of science and several departmentsof the government, 1–3. New York Bien.
Ewbank, D. C. 1987. History of black mortality and health before 1940.Milbank Quarterly 65:100–28.
Ewbank, D. C., J. C. G. de Leon, and M. A. Stoto. 1983. A reducible4-parameter system of model life-tables. Population Studies 37:105–27.
Faust, D. G. 2006. “Numbers on top of numbers”: Counting the Civil Wardead. Journal of Military History 70:995–1010.
Ferrie, J. P. 1996. A new sample of males linked from the public use mi-crodata sample of the 1850 U.S. federal census of population to the 1860U.S. federal census manuscript schedules. Historical Methods 29:141–56.
———. 2003. The rich and the dead: Socioeconomic status and mortality inthe United States, 1850–1860. In Health and Labor Force Participationover the Life Cycle: Evidence from the Past, ed. D. L. Costa, 11–50.Chicago: University of Chicago Press.
Fogel, R. W. 1986. Nutrition and the decline in mortality since 1700:Some preliminary findings. In Long-term Factors in American EconomicGrowth, ed. S. Engerman and R. Gallman, 439–527. Chicago: Universityof Chicago Press.
Glover, J. W. 1921. United States life tables, 1890, 1901, 1910, and1901–1910. Washington, DC: Government Printing Office.
Grabill, W. H., C. V. Kiser, and P. K. Whelpton. 1958. The fertility ofAmerican women. New York: Wiley.
Hacker, J. D. 1996. Adult mortality in early New England: Evidence fromthe graduates of Yale College, 1701–1805. Master’s diss. Minneapolis:University of Minnesota.
———. 1997. Trends and determinants of adult mortality in early NewEngland: Reconciling old and new evidence from the long eighteenthcentury. Social Science History 21:481–519.
———. 1999. The human cost of war: White population in the UnitedStates, 1850–1880. PhD. Thesis. Department of History. Minneapolis,University of Minnesota.
———. n.d. Rethinking the “early” decline of marital fertility in the UnitedStates. Demography 40:605–20.
Haines, M. R. 1979. The use of model life tables to estimate mortalityfor the United States in the late nineteenth century. Demography 16:289–312.
———. 1998. Estimated life tables for the United States, 1850–1910. His-torical Methods 31:149–69.
Haines, M. R., and R. C. Avery. 1980. The American life table of1830–1860: An evaluation. Journal of Interdisciplinary History 11:73–95.
Haines, M. R., L. A. Craig, and T. Weiss. 2003. The short and the dead:Nutrition, mortality, and the “antebellum puzzle” in the United States.Journal of Economic History 63:382–413.
Haines, M. R., and S. H. Preston. 1997. The use of the census to es-timate childhood mortality: Comparisons from the 1900 and 1910United States census public use samples. Historical Methods 30:77–97.
Hall, P. K., and S. Ruggles. 2004. Restless in the midst of their pros-perity: New evidence of the internal migration patterns of Americans,1850–1990. Journal of American History 91:829–46.
Harrison, R. A. 1981. Princetonians, 1776–1783: A biographical dictionary.Princeton, NJ: Princeton University Press.
Harrison, R. A., J. McLachlan, and R. L. Woodward. 1980. Princetonians,1769–1775: A biographical dictionary. Princeton, NJ: Princeton Univer-sity Press.
Hautaniemi, S. I., A. C. Swedlund, and D. L. Anderton. 1999. Mill townmortality: Consequences of industrial growth in two nineteenth-centuryNew England towns. Social Science History 1999:1–39.
Heligman, L., and J. H. Pollard. 1980. The age pattern of mortality. Journalof the Institute of Actuaries 107:49–80.
Henry, L. 1989. Men’s and women’s mortality in the past. Population: AnEnglish Selection 44:177–201.
Higgs, R. 1973. Mortality in rural America. Explorations in Economic His-tory 10:177–95.
Jacobson, P. H. 1957. An estimate of the expectation of life in the UnitedStates in 1850. Milbank Memorial Fund Quarterly 35:197–201.
Jaffe, A. J., and W. I. Lourie Jr. 1942. An abridged life table for the whitepopulation of the United States in 1830. Human Biology 14:352–71.
Johansson, S. R. 1977. Sex and death in Victorian England. In A wideningsphere: Changing roles of Victorian women, ed. M. Vicinus, 163–81.Bloomington: Indiana University Press.
Jones, E. F. 1971. Fertility decline in Australia and New Zealand 1861–1936.Population Index 37:301–38.
Kasakoff, A. B., and J. W. Adams. 1995. The effect of migration on ages atvital events: A critique of family reconstitution in historical demography.European Journal of Population 11:119–242.
———. 2000. The effects of migration, place, and occupation on adult mor-tality in the American North, 1740–1880. Historical Methods 33:115–30.
Kennedy, R. E., Jr. 1973. The Irish: Emigration, marriage, and fertility.Berkeley: University of California Press.
Kippen, R. 2005. Counting nineteenth-century maternal deaths: The case ofTasmania. Historical Methods 38:14–25.
Kunitz, S. J. 1984. Mortality change in America, 1620–1920. Human Biol-ogy 56:559–82.
Kunze, K. 1979. The effects of age composition and changes in vital rates onnineteenth century population estimates from new data. Salt Lake City,UT: Department of Economics, University of Utah.
Kuznets, S. 1965. Long swings in population growth and related economicvariables. In Economic growth and structure: Related essays, 328–78.New York: Norton.
Lee, C. 1997. Socioeconomic background, disease, and mortality amongUnion Army recruits: Implications for economic and demographic his-tory. Explorations in Economic History 34:27–55.
———. 2003. Prior exposure to disease and later health and mortality:Evidence from Civil War medical records. In Health and labor forceparticipation over the life cycle: Evidence from the past, ed. D. L. Costa,51–87. Chicago: University of Chicago Press.
Levy, D. S. 1996. Long-run geographic and temporal changes in mortalityin the colonial South. Paper delivered at the annual meeting of the SocialScience History Association, New Orleans, LA.
Logue, B. J. 1991. In pursuit of prosperity: Disease and death in aMassachusetts commercial port, 1660–1850. Journal of Social History25:309–43.
Looney, J. J., and R. L. Woodward. 1991. Princetonians, 1791–1794: Abiographical dictionary. Princeton, NJ: Princeton University Press.
April–June 2010, Volume 43, Number 2 79
McClelland, P. D., and R. J. Zeckhauser. 1982. Demographic dimensions ofthe new republic: American interregional migration, vital statistics, andmanumissions, 1800–1860. New York: Cambridge University Press.
McDaniel, A., and C. Grushka. 1995. Did Africans live longer in the ante-bellum United States? The sensitivity of mortality estimates of enslavedAfricans. Historical Methods 28:97–105.
McLachlan, J. 1976. Princetonians, 1748–1768: A biographical dictionary.Princeton, NJ: Princeton University Press.
McPherson, J. M. 1988. The battle cry of freedom: The Civil War era. NewYork: Oxford University Press.
Meech, L. 1898. System and tables of life insurance. New York: Spectator.Meeker, E. 1972. The improving health of the United States, 1850–1915.
Explorations in Economic History 9: 353–73.Meindl, R. S., and A. C. Swedlund. 1977. Secular trends in mortality in the
Connecticut Valley, 1700–1850. Human Biology 49:389–414.Murray, J. E. 1997. Standards of the present for people of the past: Height,
weight, and mortality among men of Amherst College, 1834–1949. Jour-nal of Economic History 57:585–606.
———. 2000. Marital protection and marital selection: Evidence froma historical-prospective sample of American men. Demography 37:511–21.
Neely, M. E., Jr. 2007. The Civil War and the limits of destruction. Cam-bridge, MA: Harvard University Press.
Newell, C. 1988. Methods and models in demography. New York: Guilford.Pope, C. L. 1992. Adult mortality in America before 1900: A view from
family histories. In Strategic Factors in Nineteenth Century AmericanEconomic History: A Volume to Honor Robert W. Fogel, ed. C. Goldinand H. Rockoff, 267–96. Chicago: University of Chicago Press.
Preston, S. H. 1975. The changing relation between mortality and level ofeconomic development. Population Studies 29:231–48.
———. 1976. Mortality patterns in national populations: With special ref-erence to recorded causes of death. New York: Academic Press.
Preston, S. H., and N. G. Bennett. 1983. A census-based method for esti-mating adult mortality. Population Studies 37:91–104.
Preston, S. H., and M. R. Haines. 1991. Fatal years: Child mortality inlate nineteenth-century America. Princeton, NJ: Princeton UniversityPress.
Preston, S. H., P. Heuveline, and M. Guillot. 2001. Demography: Measuringand modeling population processes. Malden, MA: Blackwell.
Ruggles, S., M. Sobek, J. T. Alexander, C. A. Fitch, R. Goeken, P. Kelly, et al.2009. Integrated Public Use Microdata Series: Version 4.0. Minneapolis,MN: Minnesota Population Center [producer and distributor].
Rutman, D. B., and A. H. Rutman. 1976. Of agues and fevers: Malaria inthe early Chesapeake. William and Mary Quarterly 33:31–60.
Schofield, R. S. 1986. Did the mothers really die? Three centuries of maternalmortality in “the world we have lost.” In The world we have gained:Histories of population and social structure, ed. L. Bonfield, R. Smith,and K. Wrightson, 229–60. Oxford: Blackwell.
Smith, D. S. 1979. The estimates of early American historical demographers:Two steps forward, one step back, what steps in the future? HistoricalMethods 12:24–38.
———. 1982. Differential mortality in the United States before 1900. Jour-nal of Interdisciplinary History 13: 735–59.
———. 2003. Seasoning, disease environment, and conditions of exposure:New York Union army regiments and soldiers. In Health and labor forceparticipation over the life cycle: Evidence from the past, ed. D. L. Costa,11–50. Chicago: University of Chicago Press.
Steckel, R. H. 1988. The health and mortality of women and children,1850–1860. Journal of Economic History 48: 333–45.
———. 1991. The quality of census data for historical inquiry: A researchagenda. Social Science History 15: 579–99.
Stolnitz, G. 1956. A century of international mortality trends: II. PopulationStudies 10: 17–42.
Sutch, R. 2006. National income and product. In Historical statistics of theUnited States, ed. R. Sutch and S. B. Carter, Volume 3, 3.3–3.6. NewYork: Cambridge University Press.
Taeuber, C., and I. B. Taeuber. 1958. The changing population of the UnitedStates. New York: Wiley.
Thompson, W. S., and P. K. Whelpton. 1933. Population trends in the UnitedStates. New York: McGraw-Hill.
Tolnay, S. E., S. N. Graham, and A. M. Guest. 1982. Own-child estimatesof U.S. white fertility, 1886–99. Historical Methods 15:127–38.
Uhlenberg, P. R. 1969. A study of cohort life cycles: Cohorts of native bornMassachusetts women, 1830–1920. Population Studies 23:407–20.
United Nations. 1982. Model life tables for developing countries. PopulationStudies no. 77. New York: United Nations.
———. 1983. Manual X: Indirect techniques for demographic estimation.New York: United Nations.
U.S. Bureau of the Census. 1921. Fourteenth census of the United Statestaken in the year 1920: Population 1920, number and distribution ofinhabitants, volume 1. Washington, DC: Department of Commerce, Gov-ernment Printing Office.
Vinovskis, M. A. 1971. The 1789 life table of Edward Wigglesworth. Journalof Economic History 31:570–90.
———. 1972. Mortality rates and trends in Massachusetts before 1860.Journal of Economic History 32:184–213.
———. 1978. The Jacobson life table of 1850: A critical re-examinationfrom a Massachusetts perspective. Journal of Interdisciplinary History8:703–24.
———. 1989. Have social historians lost the Civil War? Some preliminarydemographic speculations. Journal of American History 76:34–58.
Wells, R. V. 1992. The population of England’s colonies in America: OldEnglish or new Americans? Population Studies 46:85–102.
Wigglesworth, E. 1793. A table shewing the probability of the duration, thedecrement, and the expectation of life, in the states of Massachusetts andNew Hampshire, formed from sixty-two bills of mortality on the files ofthe American Academy of Arts and Sciences. In Vol. II, Part I of Memoirs,131–35. Boston: Andrews.
Woods, R. 1993. On the historical relationship between infant and adultmortality. Population Studies 47: 195–219.
———. 2000. The demography of Victorian England and Wales. Cam-bridge: Cambridge University Press.
Woodward, R. L., and W. F. Craven. 1991. Princetonians, 1784–1790:A biographical dictionary. Princeton, NJ: Princeton UniversityPress.
Wrigley, E. A., and R. S. Schofield. 1981/1989. The population history ofEngland, 1541–1871. New York: Press Syndicate of the University ofCambridge.
Yasuba, Y. 1962. Birth rates of the white population in the UnitedStates, 1800–1860: An economic study. Baltimore, MD: John HopkinsPress.
Zhao, Z. 2001. Chinese genealogies as a source for demographic re-search: A further assessment of their reliability. Population Studies 55:181–93.
Copyright of Historical Methods is the property of Taylor & Francis Ltd. and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.
top related