A physicist at Woods Hole: Introducing the imageintensifier and image processing to cell biology

Nancy Anderson*

Department of Visual Studies, SUNY Buffalo, USA

Feature Endeavour Vol.34 No.3

In 1963, by invitation, particle physicist George Reynolds(Princeton University) brought an image intensifier toWoods Hole Marine Biological Laboratory. Together, heand a group of biologists began experimenting with thedevice as a way to create images of cells in low-light levelsituations, especially in the study of bioluminescence. Inthis paper I am interested in how the scientists, a phy-sicist and biologists in collaboration, assessed, inter-preted and presented the pictures that they createdwith the aid of image intensification. In particular, Iconsider the problem of ‘noise’ in the image. The paperends with an example of how Reynolds and a biologist atWoods Hole contended with the presence of noise inimages used for publication. Here is an example of howdata is modified, that is, enhanced, to serve as scientificevidence. By presenting an early and simple case of thealtered image I reveal one way scientists addressed thepotentiality of presenting inappropriately modified data– a concern that has garnered much attention in thecurrent age of digital imaging technologies.

IntroductionToday, almost invariably, in cell and developmentalbiology laboratories one finds a digital camera attachedto the light microscope and cable-connected to a computerand monitor. Microscopists no longer peer down themicroscope’s tube to view their specimen set upon a stage,but rather watch the scene on an electronic screen. Thestory of how electronic imaging came to the study of livingcells, no doubt, could claim various points of origin. As forpictures of spatial resolution, beginning in 1950, televi-sion was used intermittently in biological microscopywhile at the same time, with the introduction of digitalcomputers, biologists began devising programs for auto-mating cell pattern recognition. In this paper I would liketo focus on another candidate, an instrument called theimage intensifier.When introduced to cell biologists in theearly 1960s it offered a new and real method for studyingspatially resolved images of living cells under extremelylow-light level conditions. It also presents an early case inbiology of building electronics into an experimental sys-tem and the early management of electronic image pro-duction.

Research to develop image intensifiers for night visionbecame a military priority after success in World War IIwith infra-red image detection in battlefield situations,and by the early 1950s the instrument became available,

*Tel.: +1 716 645 0544. Anderson, N. (naa6@buffalo.edu)

www.sciencedirect.com 0160-9327/$ – see front matter � 2010 Elsevier Ltd. All rights reserve

including multistage versions that increased the finalamplification through a cascading build-up of electrons(Figure 1). To begin, photons hit the image intensifier’sphotocathode causing the emission of photoelectrons thatwere then sent accelerating through electromagnetic fieldswhich kept them in spatial focus. These initial photo-electrons then hit a subsequent dynode loosening everlarger numbers of electrons. A series of collisions wasrepeated until this increased energy hit the anode andwas reconverted back into light and an image on a phos-phor screen. Among the first to use the device outside themilitary were astronomers who employed it to photographfaint stars in the far reaches of the universe, radiologistswho saw its promise for reducing patients’ exposure toradiation, and physicists who found it valuable for detect-ing high energy particles in scintillation chambers.1

Microscopists in the life sciences had always found it achallenge to study living cells as opposed to dead and fixedmaterial, and in the postwar period scientists developedand enhanced methods and techniques, e.g., ultravioletmicroscopy, phase contrast and polarizing light micro-scopy, in attempts to create detailed images of subcellularentities and events while avoiding the harmful and dis-torting effects of fixers and dyes. These techniques con-tributed immensely to improving the potential for studyingliving matter, but one of the persisting problems was tofind a way to make these observations in extremely low-light level situations. Not surprisingly, then, the imageintensifier would have drawn the attention of cell biol-ogists.

As it happened, in the early 1960s when two youngbiologists heard George Reynolds, a particle physicist onthe faculty of Princeton University, describe the imageintensifier as he used it in his own research, they proposedthat he join them, in this case, at their summerworkplace atWoodsHoleMarineBiologicalLaboratory to showthemhowthey could apply the technology to cell biology. Thus began along-termworking relationship between a physicist and thecommunity of cell biologists. Reynolds would advise andcollaboratewithbiologists on theuse of the image intensifierfrom the mid-1960s until his retirement in the late 1980s.This paper will not recount the entire history of Reynolds’biological collaborations, but will take a brief look at theearly years, circa 1965, with a special emphasis on how aphysicist joining up with a biologist could combine exper-tises in theanalysis of dataand, inparticular, theprocessingof the images they created.

1 Myron Klein, ‘Image Converters and Image Intensifiers for Military and ScientifcUse,’ Proceedings of the IRE, 47 (1959).

d. doi:10.1016/j.endeavour.2010.05.002

Figure 1. Reynolds, ‘‘Sensitivity of an Image Intensifier Film System.’’ Originally

published in Applied Optics, 5/4, April 1966: 578.

Feature Endeavour Vol.34 No.3 131

A physicist gets an invitation to Woods HoleGeorge Thomas Reynolds was born in New Jersey in 1917.As a young man he was both a boxer and a violinist, and asan undergraduate at Rutgers University he decided tobecome a physicist. Graduating from Rutgers in 1939,he enrolled at Princeton where he earned a Ph.D. inPhysics in 1943. He then spent 1 year as a Princetonpost-doctoral researcher before the Army drafted him towork at Los Alamos on the Manhattan Project where heassisted in the design of one of the atomic bomb detonators.Then, in September 1945 he traveled to Japan as oneamong the first group of scientists to assess the weapons’damages. Once released from active duty, in 1946, hereturned to Princeton and joined the physics departmentwhere he remained a faculty member until his retirementin 1987. Reynolds’ research interests involved cosmic raysand the detection of ionizing particles. In the late 1940s toearly 1950s he developed one of the first liquid scintillationcounters and in the 1950s and 1960s served as director ofPrinceton’s high energy physics program.2

Reynolds’ own introduction to the image intensifieroccurredwhilehewasona1955–1956sabbatical at ImperialCollege, London where he worked with the physicist JamesDwyer McGee, renowned expert in photoelectronic imagingand image intensifiers.3 Through his meetings with theBritish physicist Reynolds recognized the great potentialthe image intensifier could have for his own work detectinghigh energy particles. When he returned to Princeton, Rey-nolds came back with a vision of a tracking and detectionsystem with plastic filaments acting as scintillating ‘lightpipes.’ Hewould employ the image intensifier to amplify thelight in order to allow permanent photographic records ofthe particles’ pathways to be made.4

2 Hans-Jorg Rheinberger, Putting Isotopes to Work: Liquid Scintillation Counters,1950–1970. Preprint 121, MPIWG: 9–15. Republished as ‘Putting Isotopes to Work:Liquid Scintillation Counters, 1950–1970’. In J. Bernward and T. Shinn (Eds.),Instrumentation Between Science, State, and Industry (Sociology of Sciences, Vol.22). Dordrecht, Boston, London: Kluwer, 2001.

3 McGee, ‘Photoelectronic Aids in Astronomy.’ Proceedings of a Symposium onAstronomical Optics and Related Subjects. University of Manchester, Amsterdam:North Holland, 1956: 205–23; McGee, ‘Photoelectronic Image Intensifiers.’ Reports inthe Progress of Physics, 1961.

4 Nuclear and Cosmic Ray Research Aided by Development of Plastic Filaments.Princeton Herald, November 1957. Also, Reynolds, Waters, Scarl, Zdanis, ‘A Scintil-lation Chamber Image Intensifier Beam Profile Detector,’ Journal of Nuclear Instru-ments andMethods (1962): 44–8. Reynolds, ‘PhotoelectronDetection Efficiency of HighGain Image intensifier Systems Used with Scintillation Chambers’ (presented atInstrumentation for High Energy Physics Conference, CERN, July 1962) Journalof Nuclear Instruments and Methods, 1963: 226–8.

www.sciencedirect.com

In the early postwar period scintillation and Cerenkovcounters had become crucial instruments for detecting theenergy of fast moving particles. Still, it was the introduc-tion of photoelectronic amplification devices – photomul-tipliers, and then image intensifiers – that made thesystems practicable.5 These instruments detected andmagnified the radiation, converting this to photons –

enough photons to allow for photographic detection. Rey-nolds’ own professional interest in the instrument contin-ued to grow; and since the U.S. government had an on-going interest in improving this technology, Reynoldsfound himself with a long-term contract from the AtomicEnergy Commission to work with the image intensifica-tion device. Later, he would have this contract stipulatethat some funds cover his collaborative work on imageintensification with biologists.

Reynolds’ initial discussions regarding the imageintensifier with biologists occurred in the spring of1962 when Reynolds presented on physics researchand the role of electronics at a meeting of the Princetonchapter of Sigma Xi. Following his presentation two lifescientists in the audience approached him to ask aboutusing the tool for microscopic studies of bioluminescence.By the end of the conversation the two men, enthused bywhat they heard, had invited Reynolds to join them atWoods Hole that summer to begin experiments.6 Signifi-cantly, these two biologists were Shinya Inoue who held adoctorate degree from Princeton and at that time chairedthe department of anatomy and cytology at DartmouthMedical School and Robert Day Allen, who in 1962was on the Princeton biology faculty. Both men wouldgo on to develop, individually, video microscopy systemsduring the 1970s and 1980s, becoming prominent pio-neers in developing this technology for the study of livingcells.7

Reynolds could not take Inoue and Allen up on theirinvitation that summer of 1962 as he had a previousengagement to work with the cyclotron at BrookhavenNational Laboratory, but he did travel to Woods Hole inthe summer of 1963, bringing with him an image inten-sifier. He returned to Woods Hole in 1965 and 1966, andthen by the early 1970s began spending every summerthere, even purchasing a house as a rare physicist amongthe swarm of biologists residing there from May toSeptember. From the 1960s through the 1980s, heworked on a variety of projects with different biologists,including experiments with radioisotopes, biolumines-cent assays, and crystallography. Over the years hewould publish a significant series of articles on the

5 See Peter Galison, Image and Logic: A Material Culture of Microphysics, Chicago:University of Chicago, 1997: 454–5.

6 Author’s interview with George Reynolds, February 2005.7 Allen, et al., ‘Video-Enhanced Contrast, Differential-Interference Contrast

(AVEC-DIC) microscopy: A New Method Capable of Analyzing Microtubule-RelatedMotility in the Reticulopodial Network of Allogromiaa laticollaris,’ Cell Motility, 1(1981): 291–302; Allen, et al., ‘Video-enhanced Microscopy with Computer Frame’,Journal of Microscopy, 129 (1983): 3–17; Shinya Inoue published the book, VideoMicroscopy (Plenum Press, New York) in 1986. It was revised and updated in 1997,and remains a major text for this field. Additionally, Inoue still lives in Woods Holewhere every summer he continues to provide consultation and conduct classes onvideo imaging systems. In 2003 the Japan Society for the Promotion of Scienceawarded Inoue its International Prize for Biology for his work in cell biology andvideo microscopy. Allen died in 1986.

Figure 2. Reynolds, ‘‘Image Intensification Applied to Microscope Systems.’’

Originally published in Advanced Optics in Electron Microscopy, 2 (1968): 27.

132 Feature Endeavour Vol.34 No.3

application of image intensifiers to cell biology andmicroscope systems.8

The image intensifier, light and cell biologyIn the early to mid 1960s, cell biologists at Woods Holecertainly would have been familiar with the potential ofelectronic imaging devices for their work. Beginning in theinterwar period, scientists had incorporated spectroscopyand the use of the photomultiplier, which provides agraphic trace on the screen of an oscilloscope and not animage of spatial localization, to study chemical componentsof cells by means of differing light absorption patterns,most notably absorption in the wavelength range of theultraviolet. Photomultipliers had also begun to assist inwork with fluorescence markers and radioisotopes as wellas in the study of bioluminescence. It is likely that many ofthe Woods Hole scientists would have had first handexperience with photomultipliers by the time Reynoldsarrived there in 1963. During the previous academic yearReynolds, himself, had worked with Princeton biologistswho had sought to employ the photomultiplier to measurethe light emissions of a bioluminescent substances takenfrom a crustacean and a jellyfish.9 As important as thephotomultiplier was to approximating levels of light pro-duction as a one-dimensional graph, morphological map-ping of this same light remained an important goal. Theinnovation of the image intensifier, as an electronic devicein the service of microscopy, then rested in its ability toamplify the light to levels visible to the human eye and/ordetectable by film emulsions while at the same time creat-ing a spatially resolved image, a map, of the object, show-ing precisely where the light had emitted.

The first ‘image intensifier’ team at Woods Hole con-sisted of Reynolds, Allen, Inoue, J. Woodland (Woody)Hastings (University of Illinois, Champaign-Urbana) andRoger Eckert (Syracuse University). Allen and Inoue wereinterested in using the image intensifier to extend the useof polarized light in observing living cells. Hastings andEckert studied unicellular bioluminescent organisms. But,as for testing what the image intensifier could do, the firsttask to take on was to devise a complex of visualizingtechnologies by linking the image intensifier to the micro-scope, and then placing a conventional photographic cam-era in front of the image output phosphor screen (Figure 2).At this time photographic film, not videotape, was themedium of permanent records.10

Once the group of scientists had pieced together theirhybrid apparatus, testing its efficiency uncovered variousproblems. One great point of concern was the unfortunateloss of light at points of connection in the set-up. How tomarshal the limited amount of photons emitted by a speci-men with minimal loss through this diverse line-up of

8 Reynolds, ‘Evaluation of an Image Intensifier System for Microscopic Obser-vations’, IEEE Transactions on Nuclear Science, (1964): 147–51; Reynolds, ‘Sensitivityof an Image Intensifier Film System,’Applied Optics, (1966): 577–83; Reynolds, ‘ImageIntensification Applied to Microscope Systems,’ Advanced Optics in Electron Micro-scopy, 2 (1968): 1–40; Reynolds, ‘Image Intensification Applied to Biological Problems,’Quarterly Review of Biophysics, 5 (1972): 295–347; Reynolds and Taylor, ‘ImageIntensification Applied to Light Microscopy,’ Bioscience, 30 (1980), 586–92.

9 See Johnson, Shimomura, Saiga, Gershman, Reynolds, Waters, (1962), ‘QuantumEfficiency of Cypridina luminescence, with a note on that of Aequorea,’ Journal ofCellular and Comparative Physiology, 60 (1962): 85–103.10 In 1970 Reynolds and these researchers began to use videotape for recordings.

www.sciencedirect.com

amplifiers and detectors was a key challenge. The shiftfrom one component to the next in line created a precariousgap from which already scarce photons could and didscatter and escape in significant number. Still, once theybegan experimenting with the microscope–intensifier–

camera apparatus, it was calculated that they hadachieved a 104 overall gain in illumination – from the lightemitted from the organism and amplified through thestages of the intensifier to ‘light losses involved in photo-graphing the phosphor screen output.’11 This was impress-ive considering that the photocathode only managed todetect a mere 3% of the photons emitted by the specimenand, then, at the other end amplified light dispersed in thespace between the phosphor screen and the film.

All of the variables of light loss and electron motion inthis apparatus, of course, did affect the ability to create auseful image. Initial photographs taken by this groupproved to be grainy and fairly unreadable – shape eludedthe photographic medium, because short exposure timesdid not allow for a reasonable statistical build-up of elec-trons, and then photons, through the fields and to the finaleanode and phosphor screen. Already faced with the vicis-situdes of fragile, fairly unpredictable specimens – thepoint was to create images from living matter – the frus-tration of controlling errant energy particles begged for theexpertise of the physicist. Through a series of experimentsthat tested applying different voltages and exposure timesto the system, the scientists eventually began to obtainlegible pictures.

Still, of course, evenwith a legible image, not all photonslighting on the phosphor screen and developing the film atthe end of the amplification process were proper ‘informa-tional’ photons. There was the ubiquitous problem of noise,electrons generated in the intensifier that were not relatedto the source, but agitated by heat in the system, forexample, and amplified and present as ‘light’ in the finalimage. In assessing the image, researchers found thiselectronically-added noise difficult, or even at timesimpossible, to distinguish from the desired information.

11 Allen, Inoue, Reynolds, ‘Evaluation of an Image Intensifier Tube for MicroscopicObservations,’ Biological Bulletin, 125 (1963): 388. Also see Reynolds, Allen, Eckert,Hastings, Inoue, Reynolds, ‘Application of an Image Intensifier Tube to MicroscopicObservations of Bioluminescent Cells and Visualization of Weak Radioactive SourceDistribution,’ Biological Bulletin, 125 (1963): 389.

Figure 3. ‘‘Bioluminescent Flash,’’ Science cover. 5 March 1965. Reprinted with

permission of AAAS.

Feature Endeavour Vol.34 No.3 133

One could never entirely rid the system of this (dis)infor-mation, but extensive creative efforts would go into mini-mizing it.

Experiments that first summer involved identifyingspecific biomolecules with radioisotopes and, especially,the study of bioluminescence. Imitating Reynolds ownuse of scintillators in tracking high energy particles, theWoods Hole group placed a thin scintillator over a slidewhich held a radioactively-tagged specimen. The imageintensifier was attached to the microscope to amplify thelight emitted when decayed particles collided with thescintillating sheet.12 But the work that would grow intodecades-long collaborations with various biologists wasthat which began with studies of bioluminescing dinofla-gellates, unicellular organisms that can be seen as a planeof light on the ocean’s surface in many parts of the world.By the 1960s bioluminescence studies had attracted agroup of ambitious scientists interested in biochemicalquestions of proteinmetabolism. The product here, plainly,was light, and this applied nicely to the human sense ofsight as well as lent itself to electronic technologies ofmeasurement.13

The specimen and the electronically-produced imageIn the summer of 1963 the biologist Roger Eckert was atWoods Hole studying the unicellular dinoflagellate, Nocti-luca miliaris, and he hoped that Reynolds with his imageintensifier could help him confirm the precise location ofpreviously observed individual ‘microsources’ of lightwithin the cellular space, quantify light emissions andassist in understanding the triggering mechanism of thephenomenon.

Using both fluorescence light and phase contrast micro-scopy Reynolds and Eckert employed the image intensifierto track the components of the organism’s bioluminescencesystem in order to understand both its functional andmorphological arrangement. The instrument also allowedthe biologist and the physicist to generate striking photo-graphs that provided information on the location of indi-vidual sources of light, pinpointing them to the peripheralcytoplasm.

In the winter of 1965 Eckert published his first reportson the results of his work on Noctiluca miliaris in thejournal Science.14 The two-part article did not dwell on theresearch done specifically with image intensification, butinstead emphasized Eckert’s conviction that the biolumi-nescent reaction was an all-or-none action potential –

evidence of which he had found through recordings withphotomultiplier. As for the image intensifier, Eckert

12 Reynolds called the technique ‘scintillomicroscopy.’ Reynolds, Scintillomicro-scopy. Technical Report, Princeton University, October 2, 1967, ContractAT30-1, 3406.13 In the second half of the 20th century, bioluminescence studies have contributed

significantly to the arsenal of research tools for cell and molecular biology. The fireflybioluminescence system which reacts to adenosine triphosphate (ATP) has been usedas an assay for detecting and measuring this molecular ‘currency’ of energy in cells.Aequorin, a bioluminescent protein from a jellyfish, has served as a detector of freecalcium in tissues and cells. Most notably, since the mid-1990s, Green FluorescentProtein, also originating in a jellyfish, and variants from other organisms have becomeimportant transgenic fluorescent molecular markers.14 Eckert, ‘Bioelectric Control of Bioluminescence in the Dinoflagellate Noctiluca. I.

Specific Nature of Triggering Events’ and ‘Bioelectric Control of Bioluminescence inthe Dinoflagellate Noctiluca: II. Asynchronous Flash Initiation by a PropagatedTriggering Potential,’ Science, 147 (1965): 1140–5.

www.sciencedirect.com

merely mentions that the technique had provided furtherevidence that light originated out of microsources withinthe organism. One of the images from this work, however,did appear on the journal’s cover (Figure 3).

The cover image clearly shows the dense but punctatelighting pattern on the organism’s surface suggestingmicrosources, but this picture also retains the evidenceof noise – minute specks of light dotting the background,which betray it as an organic–electronic hybrid and, tothe astute viewer, might hint at a breakdown in reliable,relevant data within the space of the cell-image itself.That is, the excess of random, unruly specks acrossthe plane could suggest to viewers that perhaps not allthe depicted lightwithin the contour of the cell are relatedto the organic source – which, of course, is true. Theinteresting fact is that there has been no attempt toeliminate what might be obvious noise from this image.It might be argued that since, in this instance, the imageostensibly plays a minor role as carrier of evidence of theauthor’s claims, this extraneous signal was not much of aconcern. Most specifically, it can be said, the photographserves as ‘cover art.’ Although it is not the purpose of thispaper to analyze the role of aesthetics in scientific ima-gery, the cover was, of course, the product of, not justscientists and science journal editors, but graphicdesigners as well. To have cleaned up the noise beyondthe borders of the specimen would have, perhaps, left toomuch empty space. This cover may be seen as a spacewhere a desire for visual balance supplanted that for

134 Feature Endeavour Vol.34 No.3

visual evidential clarity. One might conclude that thephotograph landed its cover role just because it was sucha striking picture.

Noise mattered, however, in trying to extract dataacquired through electronic devices, and thus image pro-cessing procedures and tools were actively sought.Solutions were needed. Electronics had come out of worldof physicists and engineers, and so not surprising, in themid-1960s NASA budgets and developing satellite tech-nologies brought great resources to the studying, measur-ing and recording of spectra in space. These scientists toohad to contend with a great deal of ‘noise,’ and had alreadybegun developing and implementing programs to elimin-ate this undesirable signal. Thus, it was physicists andastronomers who first presented to biologists seekingnoise-reduction techniques. In many cases, at this time,procedures entailed feeding the photographs of electroni-cally-enhanced views into computers to transform theminto digital images from which noise was extracted byaveraging areas of pixel intensity. But in the case ofReynolds and Eckert, they did not have digital technologyat hand.15

Doing something about the noise problemIn most of the publications on the study of Noctilucamiliaris involving the image intensifier Eckert cites him-self as sole author, but will credit Reynolds as a photo-grapher and, even, technical director of the experiments.16

In 1967, however, an essay co-authored by Eckert andReynolds appeared in the Journal of General Physiology.17

In this piece one can certainly cull clues as to how abiologist and a physicist, one an expert in the organic whilethe other’s expertise and focus was clearly in the realm ofthe electronic, worked together to design experimentalsystems and then produce and interpret the data.

Together, in the 1967 essay, they presented in greatdetail the material upon which they had worked, and thetext displayed a fair integration of the biological and theelectronic by considering issues and data from both pointsof view. For example, their system for counting the numberof photons emitted in a flash entailed counting the numberof developed grains in a photograph and multiplying thatby the number of photons necessary to develop a singlegrain of film emulsion. When this practice rendered stran-gely inconsistent totals across specimens, speculation onthe reason fell both to organic and the electronic possibi-lities: the irregularities could reflect different nutritioncycles across Petrie dishes or the researchers may have

15 Image Processing in Biological Science (ed. Diane M. Ramsey) Berkeley and LosAngeles, University of California Press, 1968; Pictorial Pattern Recognition (Proceed-ings of Symposium on Automatic Photointerpretation (eds. GC Cheng, RS Ledley, DKPollock, A Rosenfeld), Washington D.C., 1968. See in particular, R. Nathan, ‘‘PictureEnhancement for the Moon, Mars and Man,’’ in Pictorial Pattern Recognition, 239–66.16 Reynolds is notmentioned in the reports published in the journal Science, but does

receive a photographer’s credit. Eckert, ‘‘Excitation and Luminescence in Noctilucamiliaris, Bioluminescence in Progress (F. Johnson and Y. Haneda, eds.), Princeton,1966: 269–300. In this publication Eckert provides a footnote explaining that ‘imageintensification was carried out under technical direction of, and in collaboration with,George Reynolds of Princeton University Physics Department and the Marine Bio-logical Laboratory at Woods Hole, with the support of AEC contract AT (30-1)-3406.’17 Eckert and Reynolds, ‘The Subcellular Origin of Bioluminescence in Noctiluca

miliaris,’ Journal of General Physiology, 50 (1967): 1429–58.

www.sciencedirect.com

mistakenly counted light spots as microflashes when theywere really just noise.

Photography, in particular, has held a special role incoding the concept of objectivity, even standing as themedium par excellent of ‘objective’ images. It is as if theobjects reproduce themselves on the film, one has read onmore than one occasion, or one can find various allusions tothe photograph being the ‘object itself.’ When the lightsomehow is understood as being produced and emittedfrom the object, the image may be labeled ‘autoradiog-raphy.’ In this spirit, Reynolds and Eckert called theirimage-making method autophotography to indicate thatthe organism’s own bioluminescence produced the image.But, importantly, they noted two variations on this tech-nique: direct autophotography which entailed photograph-ing the object directly from the eyepiece taking its light asit scattered out from the magnifying lenses and indirectautophotography which was a photograph taken from thephosphor screen after the light had been through itselectronic conversions via the image intensifier. As forthe attempts to represent the flashes emanating fromNoctiluca miliaris, direct autophotography failed, theauthors noted, because the light was too dim for the filmemulsions to react. The images accompanying the textwere produced by indirect autophotography.

Noise mingles with the data in these images just as itdid in the image found on the earlier picture gracing thecover of Science, but this time the authors activelyattended to the problem. Their solution was an early,and quite low-tech, example of image ‘processing,’ butthe public presentation of the altered data speaks clearlyof the epistemological and ethical problem of amendingphotographic data and the active insertion of the scientist’sexpert judgment.18 As for the preparation of results, in-cluding images, for publication and peer-review, Reynoldsand Eckert had to and did rely on pooling their expertises,biological and electronic to ‘clean-up’ their pictures, that is,eliminate this noise. How did they do it? Well, they took apen and manually inked out spots in a photograph thatthey deemed unrelated to the light generated by the organ-ism.

So how did Eckert and Reynolds settle on what to mask?When it came to these images of the flashing dinoflagellate,decisions concerning how to confront the problem of infor-mation versus noise, at least by their own admission, reliedon judgmental interpretation of the image itself and bycomparison with other images. One of their main aids wasthe overlapping of maps, the map of light taken with thehelp of the image intensifier could be superimposed on aphotograph taken through phase contrast microscopy thatrevealed locations of organelles within the cell. Spots oflight in the ‘autophotograph’ could, presumably, bematched to these organelles. Error was possible, of course,if a speck of noise randomly lined up with an organelle.

The researchers also used the criterion of size, identify-ing the smaller spots as noise. But this, of course, was not

18 Lorraine Daston and Peter Galison discuss the rise of an ethic and epistemologythat embraces the ‘trained judgement’ of the scientist. Armed with such expertise, thescientist can then authoritatively interpret, recognize resemblances and patterns andeven edit images of nature, scientific data. Daston andGalison,Objectivity. MIT Press,2007.

Figure 4. � Eckert and Reynolds, ‘‘The Subcellular Origin of Bioluminescence in Noctiluca miliaris.’’ Originally published in Journal of General Physiology, 50 (1967): 1443

(1429–1458).

Feature Endeavour Vol.34 No.3 135

foolproof for smaller bits of illumination could indicateflashes occurring on the far side of their three-dimensionalcell. Their sense of judgment, then, ran into the problem ofimperfectly reading three dimensions into a two-dimen-sional image, and they admitted that their uncertainty asto how to read perspective could cause them to misinter-pret actual biological data as ‘noise.’ In fact, their publishedconfession of possible error is one of the most fascinatingaspects of the essay. But early articles on new imageprocessing techniques could often warn of the dangers ofruining the data by taking such steps to modify the rawdata.19 So Reynolds and Eckert say upfront that theymight be fallible in that they could have mistakenly erasedsome of the smaller microsources (of light) as noise.

As noise and desired signal are essentially the samematter – electrons introduced finally to the anode andconverted into light, they had to confront the dissolutionof such a binary as signal and noise: there are points in theimage that they refer to as ‘equivocal signal’ and admit tohaving made a choice. Regaining their authority, or givingreason to accept their decisions, is their declaration thatthey provide what they believe is the ‘unequivocal pictureof the microsources.’20 But uncertainty regarding theiralteration of the image remains embedded in the essayin that the authors offer, side by side, both versions of theautophotograph, the raw ‘original’ image and the oneprocessed by Eckert and Reynolds with black ink set nextto the image of the same field taken from the phasecontrast microscope (Figure 4). Dotted circles have beenplaced on the phase contrast image to designate the pos-itions of the microsources of light. So, peers are given allpossible evidence to determine where human judgment

19 For example, Klug, De Rosier, ‘Optical Filtering of Electron Micrographs: Re-construction of One-Sided Images,’ Nature, 212 (1966): 29–32. ‘It can, however, be anextremely dangerous technique if an incorrect image is produced as a result of thewrong selection of diffracted rays for consideration.’ p. 32.20 Eckert and Reynolds.

www.sciencedirect.com

and, perhaps, ‘subjective’ interpretation might be mislead-ing.

Conclusion: the ethics of image processing todayAs the 21st century dawned digital imaging processing hadbecome a highly sophisticated norm in laboratory practice.Certainly by 2005 the fear of over-processing scientific datahad become a serious topic of discussion. Misconduct casesinvolving the processing and enhancement of scientificimages were increasing, and although the purpose wasalways claimed to be motivated by the wish for accuracyand legibility, aesthetics has also been cited as an import-ant factor in publication. But these days enhancing thedata, for whatever reason, is more andmore often seen as aproblem ready to blow up into a scandal. The Journal ofCell Biology, for example, hired an editor some years agospecifically to inspect submitted images for signs of inap-propriate modification. If there is suspicion of over-manip-ulation of the data, the managing editor will ask to see theoriginal image data.21 As careful as Reynolds and Eckertwere in their alteration of data and its public presentation(and, as they admit, there were already problems), sub-sequent digital developments in image processing arenascent in this earlier analog technology.

AcknowledgementsI would like to acknowledge the support of the Max Planck Institute forthe History of Science, Berlin, the Institute for the History of Health andMedicine, University of Geneva, and the Swiss National ScienceFoundation during the exploration of this topic and electronic imagingin cell biology in general. Also, I would like to thank the biologists as wellas historians of science and art for their comments and suggestionsduring a presentation of this material at Dartmouth College in 2006.

21 Helen Pearson, ‘Image Manipulation: CSI: Cell Biology,’ Nature, 434 (2005): 952–

3. Mike Rossner and Kenneth M. Yamada, ‘What’s in a Picture? The Temptation ofImage Manipulation,’ Journal of Cell Biology, 166 (2004): 11–5.