early mendelism and the subversion of taxonomy: epistemological obstacles as institutions
TRANSCRIPT
1
Staffan Müller-Wille, “Early Mendelism and the subversion of taxonomy: Epistemological
obstacles as institutions,” Studies in History and Philosophy of Biological and Biomedical
Sciences, 2005, 36 (3), pp. 465–487.
Last version of manuscript submitted before proofreading.
2
Abstract
The paper will present and discuss a series of hybridization experiments carried out by Nils
Herman Nilsson-Ehle between 1900 and 1907 at the plant breeding station in Svalöf (Sweden).
Since the late 1880ies, this station had been renowned for its “scientific” breeding methods,
which basically consisted in an elaborate system of keeping records by which the offspring of
individual plants was traced over generations while being meticulously described. This record
system corresponded to a certain breeding technique (the so called “Pedigree method”) and
certain theoretical convictions (mutational rise of new varieties, and non-productivity of
selection). Inspired by Tschermack's translation of Mendel's Pisum-paper, Nilsson-Ehle began
his experiments in 1900 and published a first, major synthesis of his findings in 1908. If one
compares these experiments, as they were documented in the breeding records, with their
representations in print, one encounters discrepancies in terms of procedure and presentation of
data. This can be explained by the fact, that Nilsson-Ehle was obliged to follow the recording
and breeding procedures institutionalised at Svalöf, but that these procedures, grounded in a
taxonomic discourse, left only little room for Mendelian hybridisation experiments. The twists
and turns that this story takes will be analysed in terms of Bachelardian philosophy of science,
where the “epistemological obstacle” –understood in a normative sense, as a truth, namely, that
turns out to be an error as science progresses– functions as a central, analytic category. In
contrast to Bachelard, however, I will characterise these obstacles as being of an institutional,
rather than mental nature. Thus characterized, moreover, they turn out to have been prerequisites
as much as barriers to scientific progress.
Introduction
Speaking of the ‘triumph’ of genetics in the 20th century, the ‘century of the gene’ as it is has
been called,1 has become common parlance. Depending on the historical perspective, this
3
triumphal history appears differently: Viewed retrospectively and from the standpoint of biology
as a whole, it appears as if the triumph of genetics consisted in its successful adaptation and
application to the various purposes of biological disciplines, including the overall ‘reduction’ of
biology to physico-chemical and statistical laws. The evolutionary synthesis of the 1930ies and
40ies has indeed often been described as such a success story.2 A different picture emerges,
when we look at the same story from the standpoint of genetics itself and prospectively: In 1900,
when the Mendelian laws were (re-)discovered, the aim was not to put biology as a whole on a
new footing. Classical genetics arose in a specialised research field constituted by an
autonomous, experimental basis whose core was made up by the artificial crossing of organisms,
very often carried out with specific economical aims in mind.3 From here, genetics had to
conquer its terrain, against considerable resistance from traditional biological disciplines. Some
of its basic assumptions —e.g. that developmental dispositions are transmitted independently of
each other— conflicted with well entrenched intuitions of traditional biological disciplines like
systematics or physiology.
To understand this process, epistemological categories as ‘reduction’ or ‘synthesis’ are
insufficient. They may certainly be suitable to express the personal motivation of particular
individuals active in the field. But the process as a whole has much more resemblance with a
‘subversive’ movement, a movement through which the concepts of biology —central concepts
like that of the species or the individual— were thoroughly turned inside out. Rather than a
looking for a foundational movement, we should expect from the early history of genetics the
peculiar interplay of epistemological ‘obstacles’ and ‘breaks’, which was analyzed by the French
philosopher of science Gaston Bachelard in his La formation de l'esprit scientifique (1938).4
Focussing on what he believed to be a pattern of historical progress that was peculiar to science,
Bachelard described scientific progress in his later book La philosophie du non (1940) as
follows:
4
For the scientist knowledge arises from ignorance as light from darkness. The
scientist does not see that ignorance is a fabric of positive, tenacious, coherent errors.
He does not recognise that spiritual darkness has a structure and that, under this
condition, all objective, correct experience must always determine the correction of a
subjective error. But one does not easily destroy errors one by one. They are
coordinated. The scientific mind cannot constitute itself but by destroying the non-
scientific mind.5
Here we see Bachelard chosing ‘error (erreur)’ as a central category to capture the nature
of scientific progress for the sake of its normative connotations. In contrast to what the
positivists, or ‘scientists’ (savants) as Bachelard puts it, believe, scientific progress does not
come about by the accumulation of facts, which fill, so to speak, a previous void of non-
knowledge. That is, what is in place in a field in which science makes progress, is not a lack of
knowledge, falsity in the sense of the mere absence of positive truth. What is in place is rather
error as a ‘positive, confirmed, coherent’ knowledge, which has a certain, co-ordinated structure,
and which is, or rather, has to be, destroyed by scientific progress. Being wrong in science is
therefore not simply clinging to false propositions; it is being wrong for a reason, and error in
this sense will not simply be the opposite of truth, but rather the truth that turns out to have been
an error while science progresses. Error in this sense, has a normative dimension as it is
measured against a certain standard —a standard, to be sure, that is not eternal but only
established once a progress in science is made. This is not relativism but a clear realisation of
what characterises scientific progress as an open, historical, and above all irreversible process.
Science, in short, is “the consciousness of a mind that constitutes itself [...] while searching
reality for what contradicts previous knowledge (la conscience d'un esprit qui se fonde [...] en
cherchant dans le réel ce qui contredit des connaissances anterieures).”6 Science is intrinsically
assymetrical.
5
In the following, I want to explore Bachelard’s concept of scientific progress for a
specific case. The case chosen concerns breeding experiments carried out between 1892 and
1907 at the Swedish Seed Association's experimental station in Svalöf, in the South of Sweden. I
chose this institution for two reasons, which make it probable that the establishment of the a
scientific discourse of genetics within an already existing biological discourse can especially
well be observed in this case: First, because Svalöf was one of the few institutions that carried
out breeding experiments successfully across the big divide demarcated by the rediscovery of
Mendelian rules in 1900. And second, because since 1900 a researcher, Nils Hermann Nilsson-
Ehle, worked at this institution who is known for the resolution of one of the many apparent
exceptions to the Mendelian rules, namely the inheritance of continuously varying, or fluctuating
characters. My emphasis in discussing this case will be twofold: First, I want to show, that the
Bachelardian ‘obstacles’, against which Mendelian genetics had to prevail, were not in the first
place of a mental nature (as Bachelard had it) but rather institutional. And secondly, I want to
argue that they operated in a manifold of directions: It is not only that what Nilsson-Ehle's
predecessor (and superior) at Svalöf, Nils Hjalmar Nilsson, held to be true turned out to be
erroneous by the progress Nilsson-Ehle made. It is also that Nilsson-Ehle, from the vantage point
of Nilsson, was doing the wrong thing when he set out with his Mendelian exepriments. And,
finally, that Nilsson-Ehle himself, in trying to overcome the epistemological obstacles he met
with, was not able to fully realise the ideal of Mendelian experimentation. In a certain sense,
even, one can say that Nilsson-Ehles work was crucially dependent on that done by Nilsson.
Obstacles, taken in these various dimensions, may be regarded to highlight the lines of
resistance, which, like fault-lines in a geological formation, constitute the internal history of a
scientific truth.
6
The Svalöf Method
The experimental breeding station at Svalöf was installed from resources of the Swedish Seed
Association, to whose foundation private entrepreneurs, state officials, and agricultural co-
operatives had contributed in 1886. The motive for its foundation was, that land reform and
mechanisation for the first time in Swedish history had created surpluses in agriculture,
especially in oats, that made export economically interesting. However, the English and German
cultivars that had been imported to Sweden for their high returns did not endure its winter very
well — in contrast to the traditionally cultivated, but less yielding, so-called ‘country sorts’ of
Sweden. The expressed concern of the Swedish Seed Association was therefore to test the
viability of foreign, imported seed material under the climatic conditions of Sweden, to raise its
viability, if necessary, by breeding, and to distribute the material thus tested and ameliorated to
the market.7
In accordance with this practical orientation a German agricultural engineer, Thomas
Bruun von Neergaard, was employed already in 1886. Neergard did not publish on his work, but
an impression of his procedures can be gained from his annual reports to the Association and
from descriptions of Nils Hjalmar Nilsson, who succeeded him on the post of the experimental
station's director in 1890. The breeding method Neergard used is known as ‘mass selection’ and
consisted in collecting the best plants, ears, or seeds from the harvest according to certain criteria
for quality, and to sow them out collectively for next year's harvest.8 The procedure focuses
directly on the quality that is aspired (e.g. large grains) and its effectiveness seems to be
intuitively evident. The ‘invisible hand’, that operates here, is that of the breeder himself, as it
had been active since ages, and the knowledge it is based upon, seems largely to be implicit.
Yet a closer look at Neergard's method, as it was represented by his successor Nilsson in a table,
reveals why it could, despite its intuitive foundations, put a claim on being ‘scientific’ or
‘systematic’, a claim that was raised from its very onset. The table presents a hierarchy of
7
criteria, according to which selection was carried out.9 Two observations can be made about
these criteria:
1. Selection is not carried out according to the overall impression of the plants —their
‘vigour’ or ‘beauty’— but according to certain characters, each regarded separately, and all
successively focussing step by step —first characters of the whole plant, as “structure of the
straw”, next of the ear, as weight, size, and density, and finally of the grain, as position and
weight— on the aspired quality, the quality of the grains (kärnor). Moreover, the criteria consist
of well quantifiable or classifiable characters, as weight, sizes, angles or positions. The overall
‘Gestalt’ of the plant is atomised into smallest units of measurable difference.
2. As Neergard himself noted in one of his reports, the plant material went ‘through many
hands’10. The great mass of material —in the winter of 1899/90 alone, 11000 individual plants of
a single sort of barley were examined— and the repetitive pattern of the selection process made
it necessary to employ untrained assistants, mostly women and children. According to Neergard,
his main aim lay in the factory-like organisation of this work and the mechanisation of each
working step by specially developed tools with such names as ‘combined barley- and oats-
forceps (Kombinierte Gersten- und Haferzange)’, ‘classificator (Klassifikator)’, ‘Diaphanoskop’,
and ‘earlet-sorter (Ährchensortierer)’.
This breeding method of Neergard, called ‘systematic breeding’ or simply ‘the Svalöf-
method’, was continued in its main lines by Nils Hjalmar Nilsson, who repeatedly emphasised
this in the reports and articles he published in the Swedish Seed Association's journal (founded
in 1891). And yet he maintained in 1892, when he described Neergard’s methods in an essay on
“new directions in the breeding work at Svalöf”, that the latter had only carried out “preparatory
work for the development and control of a systematic breeding method”, and that an “important
innovation” had been introduced since then.11 This innovation is apparent in the scheme
provided by Nilsson to explain Neergard’s method: The category “botanical, physical characters
8
(botaniska, fysikaliska karakterer)” is inserted into the list in normal type instead of italics. What
does it refer to?
To answer this, Nilsson's academic background has to be taken into consideration. Other
than Neeergard, Nilsson had received an education in academic botany, which can explain, why
in his account of the ‘Svalöf method’ a certain way of seeing things is prominent. After having
praised his predecessor for having introduced the ‘classificator’, an instrument for determining
the ‘density’ of the ears of cereals in a single working step without complicated mathematical
operations, Nilsson adds a long discussion of this coefficient, in which a correlation with the
‘strength of the stem’, the grain number per 100 mm of spindle length, and the ear number per
100 mm of spindle length is maintained. The coefficient of ‘density’ can thus provide the plant
breeder with a means to regard such inner developmental dispositions (innere Bildungsanlagen)
which until now lay outside his reach.” With ‘inner developmental dispositions’, to be sure,
Nilsson was not referring to something like Mendelian factors or genes, but to simple
morphological ‘laws (Grundgesetze)’ as the dependence of the density of the ears on the number
of grains per ears —dependencies which, according to Nilsson, “characterise the structural type
(Bautyp)” of a plant.12
This kind of reasoning was not really ‘new’, but a reasoning with which a botanist was
thoroughly acquainted at the end of the 19th century: It lay at the ground of taxonomically
distinguishing plant species or varieties according to their morphological type, which, in its turn,
was believed to be determined by (quasi-mechanical) laws of growth.13 But its application
prompted Nilsson to put the Svalöf method on a completely new footing with the so-called
‘pedigree method’.14 The “principle, that only newer and more noble sorts should be susceptible
to breeding” —a principle followed by Neergard quite naturally in mass selection— “has to be
thrown overboard as scientifically untenable”. Instead of concentrating on the breeding aim
itself, attention had to be directed in the first place upon the ‘variety’ of strains, taxonomically
9
distinguishable within each species of cereals, and originating from the “potential (förmåga) of
plants for spontaneous change in form (frivillig formförändring)”.15 From this variety of forms
the most promising strains could be selected on the basis of comparative tests and then subjected
to the breeding process, which, however, did not aim at changing these strains by selection, but
to ‘purify’ them from admixtures of other, co-existing strains. Heredity, understood as an
inherent, natural tendency in plants to bring forth descendants similar to their parents, was now
the breeding aim instead of certain qualities. The latter were left to an equally natural and
inherent propensity of plants for spontaneous variation —a ‘lottery’ as Nilsson expressed it.16
Gains from this lottery, however, were not left to mere chance; instead, Nilsson installed a
system of collection and record keeping at Svalöf that would maximise chances for such gains.
Keeping Records
In correspondence with his notions of heredity and spontaneous variation, Nilsson saw his chief
task in establishing and organising a ‘type collection (typsamling)’.17 Already in the first two
years of his appointment, Nilsson raised the number of cultivated cereal strains to 2000, each of
these strains occupying its own, little parcel on the experimental station's acres.18 That Nilsson
spoke of a ‘lottery’ in respect to the gains that could be expected from this variety already
evinces, that this was something that one had to be able to afford.
And indeed, one of the preconditions for changing to Nilsson's method had been a deep
institutional reorganisation of the Swedish Seed Association. As already mentioned, the
association had, at its foundation, also had the task to distribute the grains it tested for good to
the market. This had had the consequence that a large part of the area at the experimental station
was used for the mass reproduction of seed material. The Association did, in effect, not do
anything else than any other commercial plant breeder. It was only with the foundation of a
separate and independent joint-stock company in March 1891 —which alone should be
responsible for the selling of seed raised at the experimental station at Svalöf, and which, as a
10
matter of course, had exclusive access to this material— that the necessary space was opened to
realise the breeding method Nilsson envisioned. The abstraction from the ameliorisation and
proliferation of strains, inherent to the breeding principles formulated by Nilsson, had to be
realised institutionally as well to implement his pedigree method at Svalöf.19 Correspondingly,
Neergard's principles could only be ‘thrown over board’ after that institutional change had come
about. As a matter of fact, Neergard left his directorship in protest against this change, which he
viewed as going against the interest of commercial plant breeders and farmers, the clients he felt
himself obliged to.20
There was another institutional change occurring at Svalöf under the transition to the
pedigree method: While Neergard could be content with proving that he had achieved yield
increases in each of the annual reports he delivered to the Association, Nilsson's method was in
need of records that allowed to observe the development of individual pedigrees comparatively
and over successive generations. This was provided by a system of records that consisted of
three elements21: a ‘journal of analysis’, that recorded the results of comparative tests for yield
potential of different strains; ‘descent cards (härstamningskort)’, i. e. annual lists of which strains
had been cultivated on which parcels; and finally, ‘field books (fältböcker)’, in which a number
of observations on each cultivated strain was put down for each year. The latter two together
were called ‘register (stambok)’, in analogy to a family register. The journal of analysis was
mainly used to decide which strains were to be passed over to the joint-stock company for
marketing, and is not so relevant for my purposes here. A closer look at the register is in place,
however, to better understand Nilsson's breeding method.
An example for the descent cards, stemming from the year 1892 in which the register was
started, is shown in fig. 1. Its heading states the grain species (in this case oats, “hafre”), the year
(“1892”), the number of the parcel it refers to (“B.I.e. ff. [försöksfält = experimental parcel]
31”), and the oat sort that is cultivated on that parcel (black, Californian plume-oat, “Svart,
11
kalifornisk plym-”). The number for the parcel results from a complicated classification of fields
and parcels according to species sown out and the season of sowing.22 Under this heading, a list
of 10 numbered entries follows, each designating a strain distinguished on the parcel according
to botanical criteria, as the colour of the seed, the form of the ear etc. These botanical
characterisations are performed comparatively, i.e. by a listing of differences, not by independent
description, as is clear from formulations like “4. Like 3 but later and without any bristles (4.
Som 3 men senare och fullst. borstlös).”
If one takes a look at the descent cards from the following years, the way in which strains
can be followed from year to year (generation to generation) becomes clear. The original lists
from 1892 become tables, whose columns contain, from left to right, the parcel numbers (their
number rising continually), a running number (whose function is unclear to me and which later
records lack), a designation for the ‘origin (Ursprung)’ in 1892, the number of seeds sown out on
the parcel, and finally observations on seed colour, time of ripening a. s. o. The ‘origin’ is
accounted for by the number of the parcel on which the strain had been grown in 1892 (as ‘B.I.e
ff. 31) plus the number for the particular individual variety, whose seeds were sown out in 1893.
In the following years, then, it is only necessary to correlate the parcel numbers of the current
year with the parcel numbers of the previous year (fig. 2, columns 1 to 4) to retrace the plants
growing on each parcel in each successive year to their ‘origin’ from individual plants in 1892
(designated as “Pedigr.” in column 5 on fig 2). As in each year deviating botanical characters of
the plants on each parcel are carefully noted (under “Karaktärer” on fig 2), it becomes possible to
observe their behaviour —stable or variable— in the course of generations.23
In the beginning, botanical observations were inserted into the descent cards directly. But
they soon found their own place in so-called field books, which were correlated with the descent
cards by the parcel numbers. Thus a lot more space was created for these observations. I will
later discuss these entries in some more detail. A further innovation consisted in the introduction
12
of ‘register numbers (stamboksnummer)’. On figure 2 —which, by the way, shows with its mass
of ticks and underscores how intensely the descent cards were reworked in later years— they are
visible as obliquely inserted, underlined four-digit numbers in the column headed by “1894”.
These numbers were assigned to strains if these had proved to be somewhat stable, and when
they seemed promising in terms of yields. Moreover, they remained the same in contrast to the
parcel numbers, which changed annually as the plants were moved from one parcel to the other
to avoid the leaching of the soil and the production of high or low yields by coincidental
circumstances. Register numbers thus designated strains as stable, and effectively immutable
units circulating within the well-developed system of fields and parcels on Svalöf (and beyond;
Svalöf possesed subsidiary experimental stations throughout Sweden).24 They were ‘elementary
species’, ‘independent, systematic units’, as Nilsson would designate them in an article
summarising his views on plant breeding in 1907. In a sense, that is, they were atoms of the
taxonomic discourse, and correspondingly, as species and varieties in systematic botany, they
received immutable names in form of their register numbers. The breeder could only chose
among them, a slow and continuous transition from one to the other, as presupposed by mass
selection, had to appear as impossible.25
Mendelism's ascent at Svalöf
From a modern, genetically informed standpoint, one cannot easily say that the ‘Svalöf method’
as developed and systematised by Nilsson, was based on wrong theoretical suppositions. Rather,
a different, non-genetic theoretical framework was in place, that of botanical taxonomy
combined with a view of heredity and variety as fundamental tendencies or forces of life, which
was characteristic of nineteenth-century evolutionary thought. Distinguishing ‘elementary
species’ taxonomically makes sense, no matter what its genetic foundations may be, and heredity
and variation can, on an empirical level, be characterised as fundamental tendencies in the
evolution of organisms, no matter what specific mechanisms bring them about. The relation of
13
the theoretical framework informing Nilsson's breeding method to that of genetics is not so much
a matter of logical opposition, but rather one of operating on different levels of empirical
analysis.
If a relation exists between Nilsson's theoretical assumptions and later genetics, it is a
historical one that may be characterised as stopping short, as a sort of reluctance to draw
consequences. Jean Gayon has shown, that the kind of mutationism that Nilsson clung to
provided an important context for the rediscovery of the Mendelian rules in 1900.26 As a matter
of fact, Hugo de Vries, one of Mendel's rediscoverers, based his chapter on the inefficiency of
selection in his Die Mutationstheorie (1901-1903) on information he had drawn from breeder's
registers, including that of Svalöf, and he would later claim that the practical successes of the
Svalöf method approved of his theory.27 And in 1903 Wilhelm Johannsen, the creator of the
genotype-phenotype distinction, directed a questionnaire to Svalöf, asking if the ‘pure strains’
produced there were still liable to changes through selection. What Nilsson, inadvertently, had
produced in the collection of strains at Svalöf, were the ideal objects for Mendelian hybridisation
experiments, ‘pure lines’ in the sense that Johannsen explored to underpin his distinction of
phenotype and genotype in his famous 1903 booklet Über Erblichkeit in Populationen und in
reinen Linien.28 Yet Nilsson should never realise this potential. For him, hybridisation remained
a source of incalculable destabilisation of the strains he had ‘purified’ so carefully —and this
even although one of his collaborators, Per Bohlin, had identified Mendelian ratios in an
artificial cross in 1886 (without, however, creating the explanatory framework for these ratios
that Mendel provided). In 1903, Nilsson left it to one of his young collaborators, Hermann
Nilsson-Ehle, to answer Johannsen’s questionnaire, who by then had already embarked on
Mendelism.
Nilsson-Ehle had come to Svalöf in 1900 and was to be in charge of breeding
experiments with respect to oats and wheat. He was one of several assistants that had entered
14
Svalöf as a result of another series of institutional changes that happened during the late 1890ies
and by which the experimental station was divided up into several departments, each specialising
on one or two particular species of agricultural plants. Breeding experiments in Svalöf were
diversified and gained more depth, so to speak, in a hierarchy extending from the director, Nils
Hjalmar Nilsson, over department heads like Nilsson-Ehle and Bohlin and their assistants and,
finally, down to untrained field workers.
The personal reasons that prompted Nilsson-Ehle to engage in Mendelian experiments
almost immediately after he had begun work at Svalöf and against the reluctance of his superior,
must remain obscure.29 The event, that acquainted him with Mendelism, is known however:
Erich von Tschermak, another discoverer of Mendel's paper in 1900, had visited Svalöf in 1901
and presented a lecture on Mendelism there.30 A personal copy of Mendel's paper, as edited by
Tschermak in 1901, has survived in the library at Svalöf among the books of Nilsson-Ehle,
heavily annotated in his handwriting.31 These annotations may be used to clarify the points of
difference between Nilsson's mutationist and Nilsson-Ehle's Mendelian standpoint.
Nilsson-Ehle marked several sections of Mendel's text at the margin with a red line,
adding the abbreviation ‘cit.’, probably standing for “quote (citera)”. All of them speak directly
against Nilsson's taxonomic standpoint. The first concerns Mendel's assertion, that the taxonomic
rank of the experimental plants is irrelevant32 —Nilsson, as we saw, put great weight on an exact
taxonomic determination of his experimental plants. Secondly, Nilsson-Ehle marked Mendel's
formulation of the law of independent assortment, according to which “the constant characters,
which appear in the several varieties of a group of plants may be obtained in all the associations
which are possible according to the [mathematical] laws of combination, by means of repeated
artificial fertilisation.”33 Nilsson had not only maintained the interdependency of characters, but
had even built his method upon morphological ‘laws’ expressing such interdependencies. And,
finally, Nilsson-Ehle marked a whole page on which Mendel discussed the view that cultivation
15
entails destabilisation of the species, containing the denial of the position that “their offspring
diverge into an endless seriesof extremely variable forms”34 —again a belief, as mentioned
shortly before, that Nilsson held. The core of all three passages is the same, namely that the
entities which the taxonomist distinguishes (‘forms’), are nothing but free, uninhibited
combinations of elementary characters, combinations which therefore obey the laws of
combinatorics35 —a view that had to be wholly unintuitive to Nilsson.
Yet it would be rash to assume that Nilsson-Ehle, in contrast to Nilsson, grasped the
mathematical core of Mendel's theory immediately. The last page of his personal Mendel-copy is
filled with little sketches, making use of an abbreviated version of the scheme Mendel himself
had provided to visualise the possible combinations between pollen- and germ-cells of hybrids
(fig. 3). Successively, Nilsson-Ehle is trying to derive the combination series (and numerical
ratios) for a hybrid self-fertilisation, the two back-crosses of hybrids with their respective parent
forms, and the F2-series for a dihybrid cross. The first three series are developed, yet with some
difficulties, as evinced by corrections. However, the dihybrid-series, which is not visualised as
easily by a scheme that would be analogous to the schemes for monohybrid crosses (an attempt
in the middle of the page is struck out), resists Nilsson-Ehle's attempts. He is unable to develop
the full series of combinations, listing only a few. These difficulties in intuiting a mathematical
operation —easily resolved in taking recourse to algebra, as Mendel indeed did in his paper—
where, however, not the only difficulties Nilsson-Ehle met in his attempt to introduce Mendelian
reasoning and experimenting in Svalöf.
Breaking the limits of representation
The first outspoken formulation Nilsson-Ehle gave of his dissenting views appeared in the
Journal of the Swedish Seed Association from 1907 (alongside Nilsson's discussion of
‘elementary species’ which I mentioned above). It was entitled “North-Scandinavian and other
early oat strains and attempts to ameliorate them by individual [i.e. pedigree] breeding and
16
hybridisation” and contained the first formulations of Nilsson-Ehle's theory of inheritance by
multiple factors.
In two points Nilsson-Ehle dissented, more or less outspokenly, from his superior's
views, without, however, explicitly mentioning him: Firstly, “that it can happen, that what
appears to us as one property, may in fact be a composite (sammansättning) of several
properties”36; and, secondly, “that breeding should not proceed from the assumption, that certain,
aspired combinations [of properties] may not be achieved.”37 The first assertion contradicted the
taxonomic position, which regards visible characters as un-analyzable units (and was directly
inspired from a remark in Mendel's paper on a Phaseolus-cross that had yielded continuous
colour variation in F2); the second contradicted the taxonomic position that characters depend on
each other, precluding certain combinations of characters and thus constituting ‘types’ by their
correlation. In effect, though not naming him, Nilsson-Ehle accused his superior of being wrong
on these two essential points.
The basis for these assertions was a hybridisation experiment that Nilsson-Ehle had
carried out in previous years. This experiment began in 1903 with an artificial crossing —
artificial in as much as oats is a species subject to obligatory self-fertilization— of two strains of
oats, which had been cultivated on Svalöf for a long time already and carried the register
numbers 0353 and 0462, later changed to 0668. Fig. 4 shows a field book entry for one of these
strains, register number 0462, for the year 1903. A field book entry for one of these strains from
the year 1903 contains, as all field book entries do as a rule, the following information: the year
of the entry in the upper left hand corner; the number of the parcel from which the seeds were
derived in the previous year in the upper right hand corner; the parcel number and the register
number, followed by a detailed botanical description of the plants on the parcel; further
information on steps in the development and cultivation of the plant; and finally, the number of
the parcel on which seeds from these plants were sown out in the following year 1904. Each field
17
book entry, which, as a rule, occupies a single page of the field book, thus represents the plants
grown on one parcel as a distinct unit in terms of botanical character, development/cultivation,
and descent.
From the artificial crossing of strains 0353 and 0462, four individuals were raised in 1904
on one parcel, for which the field book entries look similar, if only differentiated for each
individual (“376a-d”, see fig. 4). As expressly stated in the entries c and d (“lik 376a”), and as to
be expected by the Mendelian rules for the first hybrid generation, the four plants are similar to
each other, except for 376b, which is said to possess slightly shorter seeds.
The seeds from the four individuals of 1904 were sown out on separate fields in the
following year. Now, however, the field book entries exhibit a completely different structure
(fig. 5): A botanical description of the plants growing on each of the parcels is lacking. Instead a
single category —the colour of the seed— is observed in regard to the absolute frequency with
which its alternatives appear on the parcels (small column in the right hand middle of the page):
“8 grey (gråa)”, “4 white (hvita)”, “207 black (svarta)”, in various shadings (“mörkare”,
“ljusare”). Contrary to Mendelian expectations, colour did not segregate discretely, and not at all
in a 3:1 proportion.
It is these absolute frequencies, tabulated in a little diagram for the descendants of each of
the four plants raised in 1904 (“a,b,c,d”), and their strong deviation from what should be
expected from the Mendelian rules, to which Nilsson-Ehle's 1907 essay refers in maintaining –
against Nils Hjalmar Nilsson – “that it can happen, that what appears to us as one property, may
in fact be a composite of several properties”.38 Why this should be so, however, is not explicitly
discussed in this essay. Only a year later, in a contribution to a professional botanical journal, did
Nilsson-Ehle interpret his findings under the assumption, that the “black colour consists of two
independent units (unabhängige Einheiten)” and that its hereditary pattern therefore follows the
“dihybrid scheme” of Mendel.39 In maintaining this, the table of 1907 is extended in three
18
respects (fig. 6). First, the absolute frequencies are transmuted into relative frequencies; second,
the sum (the mean) is drawn from the absolute (relative) frequencies; and finally, a symbolic
representation is added for the dihybrid scheme. Through this series of mathematical and
symbolical operations Nilsson-Ehle brings about an abstraction from the concrete distribution of
individuals and strains on fields and parcels that allows him to present his data as in accordance
with the Mendelian rules. The same abstraction was already implicitly present in his 1907 article,
where he conceived of the descendants of his hybridisation experiments as a ‘population’, not as
individual strains.40
How Nilsson-Ehle reached his conclusions regarding the ‘dihybrid scheme’ determining
seed colour in 1907 is made evident by a leaflet he inserted into the field-book of 1905 (fig. 7). It
develops the dihybrid scheme for seed colour by assuming two factors —one for black (“svart”),
one for grey (“grå”) colour— which in turn might either be present or absent (‘un-black’ and
‘un-grey’, “osvart” and “ogrå” in the latter case). The possible combinations of these four
alternatives in the zygote are listed under a hybridization scheme similar to the one Nilsson-Ehle
had used in his annotations to the Mendel paper and should later also use in his printed
presentation in 1908. The resultant colour of the seed is derived from these combinations under
the assumption that absence of both the factor for black and for grey yields white and that black
“covers” grey, i. e. yields the same black colour independent of the presence or absence of grey
factors. The resultant relative frequencies to be expected for black, grey, and white colour —that
is, 12:3:1— are noted on the left. The article of 1908 interprets the data under a different
assumption to bring the observed absolute frequencies into better accord with the theoretically
developed frequencies. It starts from the assumption that black colour results from two factors
for black which severally and jointly effect black colour, only their complete absence resulting in
white and greyish colours (produced by the absence of any factor for black colour).41
19
Such traces in the register evince that a separate space of symbolic representation had to
be opened by Nilsson-Ehle to explore the regime of Mendelian combinatorics in order to ‘see’
how assumed factors might combine to reproduce the empirically observed distribution of
characters. Such distributions, that is, were not simply to be ‘seen’ from what the register itself
presented as data. They were rather developed in a trial-and-error fashion by performing
mathematical permutations with symbolical representations in a separate paper space and
comparing the results with the data contained in the register.42 The recording system instituted at
Svalöf, however, did not leave much space for such manipulations. This becomes especially
clear, when we look at how Nilsson-Ehle's experiment with the cross 0353 x 0668 proceeded:
For 1907, the corresponding field-book is not preserved, and the notes on the continued
experiment contained in the 1908 field book offer a disappointment. Except for a few notes on
the cultivation and development of the plants, it only offers the following dry remark (fig. 15):
“Numbers 272-317 of cross 0353 x 0668 have been handled from a theoretical viewpoint. See
separate lists (Korsningsnumren 0353 x 0668 272-317 är bearbetad ur teoretisk synpunkt. se
särskilda listor)”.
Were these lists have ended up, is unknown. They have vanished from the archive at
Svalöf, were the other material used in this paper is otherwise exceptionally well preserved.
Some leaflets from later experiments have survived in the Nilsson-Ehle papers preserved in the
Lund University library (Nilsson-Ehle destroyed most of his papers when he became afflicted
with depression in the late 1930s). And these show that the tables and diagrams Nilsson-Ehle
produced occupied more and more space: the tables drew together counts over a widening range
of fields (see fig. 8, where field numbers are tabulated against counts for a continuously varying
character, time of germination or “groningsmognad” in days, “dagar”), and the diagrams
explored more and more complex combinations of factors (see fig. 9, were the ratios for a
tetrahybrid cross are developed).
20
But it was not only the lack of paper space that inhibited Nilsson-Ehles early experiments. The
leaflet inserted in the 1905 field-book contains a further information (fig. 7), namely which of
the combinations developed in the scheme, if left to self-fertilisation in the following
generations, should be expected to remain “constant (konstant)” and which to be “variable
(var.)” —‘constant’ meaning that no segregation into differently coloured descendants should be
expected, and ‘variable’ meaning the contrary. If these predictions about behaviour in further
generations were fulfilled, a decisive proof for the interpretation would have been reached. And
indeed, the 1908 article contains information on this point: “Of 43 plants chosen at random
(beliebig) from the second [hybrid] generation, about one half (21) turned out to be constant [in
the third hybrid generation], i. e. only resulted in descendants of black and white colour
respectively. The rest of 22 plants (with black seed colour) showed segregation black : white,
and this in various proportions [...]. Theoretically, this must be the case, too.”43 This information,
however, deviates from the 1907 report on the same experiment in a decisive respect: There it
was said explicitly, that the 43 plants had not been chosen at random, but according to “early
ripeness and formation of ear”. Nilsson-Ehle himself strongly emphasised how important it was
to renounce from any kind of selection in Mendelian analysis.44 How do we explain, then, that he
did carry out selection after all?
The essay of 1907 stated explicitly, that “there are practical limits to the possibility to
undertake [a Mendelian] analysis”. If it should “be undertaken completely and each plant should
be sown out for itself, one would get several thousand parcels already in the third [hybrid]
generation. Under a practical point of view one therefore has to limit oneself to a small number
of plants [...].”45 That Nilsson-Ehle did not choose the option, that he maintains to have chosen in
his 1908 article, has to do with the circumstance that he —as the introduction to the 1907 article
evinces— actually subscribed to the breeding aim at Svalöf, namely to combine high yields with
winter hardiness. Consequently he returned to the pedigree method in the third hybrid
21
generation, i.e. the separate cultivation of types selected for a specific, overall character (in this
case “earliness”). Not in the least instance, this is demonstrated by the field book entries for the
third hybrid generation in 1906, which return to the old pattern: For each parcel a taxonomic
characterisation of the different types (“A”, “B”, etc.) is carried out, that also, but not only, and
especially not always (namely in those cases, were it does not make a difference), records seed
colour. The segregation ratios for seed colour in the third hybrid generation have to be “distilled”
painstakingly from these 46 entries, and are anyway, with the low absolute numbers and the
previous selection, anything else than trustworthy.
Conclusion
If Nilsson-Ehle, as an employee of the experimental station, saw himself forced to return to the
standards of carrying out and recording breeding experiments instituted there, or if he hoped to
hit upon a stable, promising type combining high yields with winter hardiness in the third
generation already —in this he admittedly saw himself deceived46— is a matter of speculation.
What I wanted to show was, that the Mendelian analysis that Nilsson introduced at Svalöf very
soon encountered obstacles, which were posed by institutionalised methods of experimentation
and registration of experimental results inspired by botanical taxonomy and pedigree-breeding,
and which, if at all, Mendelian analysis could only overcome by undermining or obviating them.
These obstacles were not of a psychological nature, but were inscribed in the material
organisation of the experimental station at Svalöf itself. The parcels on the fields of Svalöf and
the register of descendant cards and field books both served the representation of types and were
organised accordingly. And as parcels and register pages were scarce resources at Svalöf —each
parcel and each field book entry consumed space, time, and work—, each spread of a new
system of representation threatened the representation of types not only as an intellectual
alternative, but in its material existence as well.
22
The erroneous views Nilsson-Ehle accused his superior of were thus not only mental entities.
They structured the material world in which he worked. In that sense, and despite their refractory
nature, they can even be said to have constituted the prerequisite upon which Nilsson-Ehle was
able to make scientific progress. The objects he experimented with where only given through the
elaborate system of register pages and isolated pedigrees that Nilsson had raised in Svalöf. And
although, Nilsson-Ehle aimed to analyze them through their hybridization, and thus to
decompose, to dismember, to ‘destroy’ them in a literal sense, his work was crucially dependent
on their existence. In Nilsson’s world, Nilsson-Ehle was progressing along a wrong, at least
dangerous way. In the world, Nilsson-Ehle was about to create —in 1925 he should succeed
Nilsson on the post of director— it would be Nilsson who had been going in the wrong direction.
Only naturally, the zone of transition between these two worlds was filled with errors,
inconsistencies, and conflicts.
23
References
Allen, Garland E. 1990. Chevaux de course et chevaux de trait. Metaphores et analogies agricoles dans l'eugenisme americain 1910-1940. In Histoire de la genetique. Pratiques, Techniques, et Theories. Paris: A. R. P. E. M. & Editions Sciences en Situation.
———. 1991. Essay Review: History of Agriculture and the Study of Heredity – a New Horizon. Journal of the History of Biology 24 (3):529-536.
Andersson, Gösta. 1986. Sveriges Utsädesförening. Historisk översikt. Sveriges Utsädesförenings Tidskrift 96 (1):7-26.
Anon. 1891. Utsädesföreningen och Utsädesbolaget. Allmänna Svenska Utsädesföreningens Tidskrift 1:24-27.
———. 1892. Vägledning genom Svenska Utsädesföreningens Försöksfält på Svalöf 1892. Allmänna Svenska Utsädesföreningens Tidskrift 2:93-110.
———. 1893. Vägledning genom Svenska Utsädesföreningens Försöksfält på Svalöf 1893. Allmänna Svenska Utsädesföreningens Tidskrift 3:80-114.
Bachelard, Gaston. 1938. La formation de l'esprit scientifique. Contribution à une psychoanalyse de la connaissance objective. Paris: Vrin.
———. 1949. Le philosophie du non. Essai d'une philosophie du nouvel esprit scientifique. 2e edition. Paris: Presses universitaires de France.
———. 2002. The formation of the scientific mind. Translated by Mary McAllester Jones. Manchester: Clinamen Pr.
Fitzgerald, Deborah. 1990. The Business of Breeding: Hybrid Corn in Illinois, 1890-1940. Ithaca, N.Y.: Cornell University Press.
Gayon, Jean. 1998. Darwinism´s Struggle for Survival. Heredity and the Hypothesis of Natural Selection. Cambridge/Mass.: Cambridge Univ. Pr.
———. 2000. From measurement to Organization: A Philosophical Scheme for the History of the Concept of Heredity. In The concept of the gene in development and evolution : historical and epistemological perspectives, edited by P. Beurton, R. Falk and H.-J. Rheinberger. Cambridge: Cambridge Univ. Press.
Gayon, Jean, and Doris T. Zallen. 1998. The role of the Vilmorin Company in the promotion and diffusion of the experimental science of heredity in France, 1840-1920. Journal of the History of Biology 31:241-262.
Harwood, Jonathan. 1993. Styles of scientific thought : the German genetics community ; 1900 - 1933, Science and its conceptual foundations. Chicago: Univ. of Chicago Press.
Keller, Evelyn Fox. 2000. The century of the gene. Cambridge, Mass. [u.a.]: Harvard University Press.
Kimmelmann, Barbara A. 1990. Agronomie et théorie de Mendel. La dynamique institutionelle et la génétique aux Etats-Unis, (1900-1925). In Histoire de la genetique. Pratiques, Techniques, et Theories. Paris: A. R. P. E. M. & Editions Sciences en Situation.
24
Klein, Ursula. 2001. Paper Tools in Experimental Cultures. Studies in the History and Philosophy of Science 32 (2):265-302.
Kloppenburg, Jack R. Jr. 1985. First the seed: A social history of plant breeding and the seed industry in the United States, Dissertation Abstracts International 46: 279-A, Univ. Microfilms order no. 85-04530., Cornell Univ.
Larsson, Bengt M.P., Mats Morell, and Janken Myrdal, eds. 1997. Agrarhistoria. Stockholm: LTs förlag.
Lenoir, Timothy. 1982. The Strategy of Life. Teleology and mechanics in nineteenth-century germany. Chicago - London: The University of Chicago Press.
Mendel, Gregor. 1901. Versuche über Pflanzenhybriden : zwei Abhandlungen, 1865 und 1869. Herausgegeben von Erich Tschermak. 7. Aufl. ed, Ostwald's Klassiker der exakten Wissenschaften. Nr. 121. Leipzig: Wilhelm Engelmann.
———. 1902. Experiments in plant-hybridisation. In Mendel's Principles of Heredity. A Defence, edited by W. Bateson. London: Cambridge University Press.
Neergaard, Th. Bruun von. 1890. Huru svenska utsädesfrågan förts på afvägar genom herr Birger Welinder från slutet af juli 1885- 2 april 1890 : Belysning af en nutidsfråga med särskild hânsyn till utsädesföreningens och sv.utsädesfrågans framtid. [Med bilaga.]. Stockholm.
Nilsson, Nils Hjalmar. 1891. Hveteförädlingens nu varande ståndpunkt på Svalöf. Allmänna Svenska Utsädesföreningens Tidskrift 1:127-133.
———. 1892. Några nya uppslag vid förädlingsarbetena på Svalöf. Allmänna Svenska Utsädesföreningens Tidskrift 2:125-135.
———. 1893. Huru draga slutledningar från olika års jemförande försök? Allmänna Svenska Utsädesföreningens Tidskrift 3:163-166.
———. 1907. De elementära arternas betydelse för växtförädlingen. Sveriges Utsädesförenings Tidskrift 17:197-208.
Nilsson-Ehle, Nils Herman. 1907. Om lifstyper och individuell variation. Botaniska notiser 1907:113-140.
———. 1907. Om nordskandinaviska och andra tidiga hafresorter och försök till deras förbättrande genom individualförädling och korsning. Sveriges Utsädesförenings Tidskrift 17:209-218.
———. 1908. Einige Ergebnisse von Kreuzungen bei Hafer und Weizen. Botaniska Notiser 1908:257-294.
———. 1924. Versuche über die Möglichkeit, Frühreife mit hoher Ertragfähigkeit durch Kreuzungen zu kombinieren. In Beiträge zum landwirtschaftlichen Pflanzenbau, insbesondere Getreidebau. Festschrift zum 70. Geburtstage Prtofessors Dr. h. c. Franz Schindler, edited by D. L. i. Brünn. Berlin: Paul Parey.
Olby, Robert C. 1985. Origins of Mendelism. 2nd ed. Chicago – London: Univ. of Chicago Pr. Original edition, 1966.
Olsson, Gösta. 1993. Sveriges Utsädesförenings tillkomst och första utveckling. Sveriges Utsädesförenings Tidskrift 103:171-190.
Palladino, Paolo. 1990. The Political Economy of Applied Research: Plant breeding in Great Britain, 1910-1940. Minerva 28:446-468.
25
Paul, Diane B., and Barbara A. Kimmelmann. 1988. Mendel in America: Theory and Practice 1900-1919. In The Amereican Development of Biolgy, edited by K. Benson, J. Maienschein and R. Rainger. Philadelphia: Univ. of Pennsylvania Pr.
Roll-Hansen, Nils. 1978. The Genotype Theory of Wilhelm Johannsen and its Relation to Plant Breeding and the Study of Evolution. Centaurus 22 (3):201-235.
———. 1986. Svalöf and the origins of classical genetics. In Svalöf 1886-1986: Research and results in plant breeding, edited by G. Olsson. Stockholm: LTs Förlag.
———. 1990. Le croisement des lignes pures: De Johannsen a Nilsson-Ehle. In Histoire de la genetique. Pratiques, Techniques, et Theories, edited by J.-L. Fischer and W. H. Schneider. Paris: A. R. P. E. M. & Editions Sciences en Situation.
———. 1997. The role of genetic theory in the success of the Svalöf research station. Sveriges Utsädesföreningens Tidskrift 107 (4):196-207.
Shireff, Patrick. 1873. Improvement of the cereals and an essay on the wheat-fly. Edinburgh - London: Blackwood.
Smocovitis, Vassiliki Betty. 1996. Unifying Biology. The Evolutionary synthesis and Evolutionary Biology. Princeton/N. J.: Princeton Univ. Pr.
Special-Katalog der Kollektiv-Ausstellung des Allgemeinen Schwedischen Saat-Zucht-Vereins Svalöf. 1890. Malmö: Allmänna Svenska Utsädesaktiebolaget.
Stubbe, Hans. 1965. Kurze Geschichte der Genetik bis zur Wiederentdeckung der Vererbungsregeln Gregor Mendels. Edited by H. Stubbe. 2 ed. Vol. Beitrag 1, Genetik. Grundlagen, Ergebnisse und Probleme in Einzeldarstellungen. Jena: VEB Fischer.
Sveriges utsädesförening 1886-1936 : en minnesskrift, ed.^eds. 1936. Malmö: Skånska litogr. ab.
Welinder, Birger. 1891. Årsberättelse för 1890. Allmänna Svenska Utsädesföreningens Tidskrift 1:3-11.
26
Endnotes
1 (Keller 2000).
2 (Smocovitis 1996).
3 (Paul and Kimmelmann 1988; Kloppenburg 1985; Allen 1990; Palladino 1990; Fitzgerald
1990; Kimmelmann 1990; Allen 1991).
4 (Bachelard 1938), ch. 1; an English translation has only recently been published in (Bachelard
2002).
5 (Bachelard 1949), 8.
6 (Bachelard 1949), 9.
7 (Larsson, Morell, and Myrdal 1997), 312.
8 (Olsson 1993), (Roll-Hansen 1990),104-106.
9 (Nilsson 1892), 126.
10 Quoted in (Olsson 1993), 181
11 (Nilsson 1892), 125, 130.
12 (Special-Katalog der Kollektiv-Ausstellung des Allgemeinen Schwedischen Saat-Zucht-Vereins
Svalöf 1890): 29-33.
13 (Lenoir 1982).
14 The phrases used by Nilsson in describing this method indicate that he received the main
impulse from (Shireff 1873); on this essay and its impact see (Stubbe 1965): 87. On the pedigree
method in general see (Gayon and Zallen 1998).
15 (Nilsson 1892), 131.
16 (Nilsson 1892), 131; cf. (Roll-Hansen 1986).
17 (Nilsson 1891), 129.
18 (Anon. 1893), 85.
27
19 (Anon. 1891): 24-27.
20 Neergard formulated his protest in a little brochure (Neergaard 1890), which he published after
he had left Svalöf.
21 Cf. (Welinder 1891): 7.
22 Vgl. (Anon. 1892): 99-100.
23 Thus some of the successfully marketed grain sorts from Svalöf can be retraced to a single ear
sown out in 1892; see (Sveriges utsädesförening 1886-1936 : en minnesskrift 1936), 54-56.
24 (Nilsson 1893). On the system of subsidiary stations of Svalöf see (Andersson 1986), 18-23.
25 (Nilsson 1907).
26 (Gayon 1998), 253ff., cf. (Gayon 2000).
27 (Roll-Hansen 1997), 203.
28 (Roll-Hansen 1978), (Roll-Hansen 1990).
29 It seems connected to the differences in thought style between ‘mandarins’ and ‘outsiders’,
which Jonathan Harwood has described for German genetics in the early twentieth century
(Harwood 1993).
30 (Roll-Hansen 1997), 203.
31 Svalöf-Weibull AB, Svalöf, library
32 (Mendel 1901), 6.
33 (Mendel 1901), 22: “[...] dass constante Merkmale, welche an verschiedenen Formen einer
Pflanzensippe vorkommen, auf dem Wege der wiederholten künstlichen Befruchtung in alle
Verbindungen treten können, welche nach den Regeln der Combination möglich sind.” English
translation quoted from (Mendel 1902), p. 65.
34 (Mendel 1901), 36: “[...] ihre Nachkommen in einer endlosen Reihe höchst veränderlicher
Formen aus einander gehen.” English translation quoted from (Mendel 1902), p. 82.
35 According to (Olby 1985), 132, this is indeed the essential characteristic of Mendelism.
28
36 (Nilsson-Ehle 1907), 214.
37 (Nilsson-Ehle 1907), 217/218.
38 (Nilsson-Ehle 1907), 214.
39 (Nilsson-Ehle 1908): 266.
40 (Nilsson-Ehle 1907): 217. In this, Nilsson-Ehle referred to Johannsen’s work. Nilsson-Ehle
had been in contact with Johannsen since at least 1903; cf. (Roll-Hansen 1990).
41 I thank Nils Roll-Hansen for drawing my attention to this discrepancy.
42 See (Klein 2001) for the significance of “paper tools” in scientific practice.
43 (Nilsson-Ehle 1908): 266.
44 (Nilsson-Ehle 1907): 114-115.
45 (Nilsson-Ehle 1907): 215.
46 See (Nilsson-Ehle 1924), where Nilsson-Ehle admits difficulties in combining early ripeness
with winterhardiness without prior “direct” selection.