Data in Time

Author(s): DAVID SEPKOSKI

Source: Historical Studies in the Natural Sciences , NOVEMBER 2018, Vol. 48, No. 5,
SPECIAL ISSUE: HISTORIES OF DATA AND THE DATABASE (NOVEMBER 2018), pp. 581-593
Published by: University of California Press

Stable URL: https://www.jstor.org/stable/10.2307/26616639

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/10.2307/26616639?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

University of California Press is collaborating with JSTOR to digitize, preserve and extend access
to Historical Studies in the Natural Sciences

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms
D A V I D S E PK O S K I*

Data in Time: Statistics, Natural History,

and the Visualization of Temporal Data

A B S TR A C T

One of the best arguments for approaching the history of information processing and
handling in the human and natural sciences as a “history of data” is that it focuses our
attention on relationships, convergences, and contingent historical developments that
can be obscured following more traditional areas of focus on individual disciplines or
technologies. This essay explores one such case of convergence in nineteenth-
century data history between empirical natural history (paleontology and botany),
bureaucratic statistics (cameralism), and contemporary historiography, arguing that
the establishment of visual conventions around the presentation of temporal patterns
in data involved interactions between ostensibly distinct knowledge traditions. This
essay is part of a special issue entitled Histories of Data and the Database edited by
Soraya de Chadarevian and Theodore M. Porter.
KEY WORDS: statistics, natural history, visualization, cameralism, data

In recent years, historians of science have been drawn to a burgeoning sub-

field, the “history of data.”1 Although there have been many good arguments
for adopting data as a historical category, there are legitimate questions about
how this recent scholarly interest relates to established histories of quantifica-
tion, statistics, mathematics, computing, and of the various disciplines in the
natural and social sciences. My own position is that histories of data do not
replace or subordinate these other research programs, but rather bring together
scholarly perspectives that have not otherwise necessarily been in conversation

*Department of History, University of Illinois, Urbana-Champaign, 301 Gregory Hall, 810 S.

Wright St., Urbana, IL 61801, [email protected]
1. See, for example, Elena Aronova, Christine von Oertzen, and David Sepkoski, eds., “Data
Histories,” Osiris 32 (2017).

Historical Studies in the Natural Sciences, Vol. 48, Number 5, pps. 581–593. ISSN 1939-1811,
electronic ISSN 1939-182X. © 2018 by the Regents of the University of California. All rights
reserved. Please direct all requests for permission to photocopy or reproduce article content
through the University of California Press’s Reprints and Permissions web page, http://www.
ucpress.edu/journals.php?p¼reprints. DOI: https://doi.org/10.1525/hsns.2018.48.5.581.

| 581

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms
582 | SEPKOSKI

to shed light on practices, epistemologies, technologies, and cultures around

the collection and analysis of data in the longue durée. This has been my
approach in recent work linking nineteenth-century natural history, state
administrative sciences, and humanistic historical conventions, where I have
found that it has been illuminating, for example, to link the work of Ted Porter
and others on the history of statistics with studies of natural history and
geology in the tradition of Martin Rudwick.2
This essay does not attempt to summarize those more detailed studies, but
rather offers some broad reflections on a particular theme in the history of data:
the turn toward visual depictions of temporal patterns in data in the
nineteenth-century natural and economic sciences. It has long been established
that the nineteenth century was a period when statistical arguments became
a central component of scientific epistemology in Europe and beyond. This
was a time during which, to use Porter’s memorable phrase, social and natural
scientists established “trust in numbers” as the basis for statistical decision-
making in a wide variety of disciplines.3 This was a process that was political as
well as methodological and epistemological: statistical reasoning, as Porter,
Alain Desrosiers, Ian Hacking, and others have demonstrated, requires the
tacit social approval that numbers and statistical objects support valid repre-
sentations of reality, and that their analysis can provide a sound basis for
economic and social interventions.4 The statistical turn introduced the broad
acceptance of a new kind of object—the statistical aggregate—that was neither
wholly abstract, being derived from individual data points, nor reducible to
any specific individual or moment in time, but able to stand for a general class.
Statistical entities such as “the average man,” “unemployment,” or “crime rate”

2. David Sepkoski, “Towards ‘a Natural History of Data’: Evolving Practices and Epis-
temologies of Data in Paleontology, 1800–2000,” Journal of the History of Biology 46, no. 3 (2013):
401–44. Sepkoski, “The Earth as Archive,” in Archiving Sciences: Pasts, Presents, Futures, ed.
Lorraine Daston (Chicago: University of Chicago Press, 2017), 53–83; David Sepkoski and Marco
Tamborini, “‘An Image of Science’: Cameralism, Natural History, and the Visual Language of
Statistics in the Nineteenth Century,” Historical Studies in the Natural Sciences 48, no. 1 (2018):
56–109. Theodore M. Porter, Trust in Numbers: The Pursuit of Objectivity in Science and Public
Life (Princeton. NJ: Princeton University Press, 1995); Porter, The Rise of Statistical Thinking,
1820–1900 (Princeton, NJ: Princeton University Press, 1986). Martin J. S. Rudwick, Bursting the
Limits of Time: The Reconstruction of Geohistory in the Age of Revolution (Chicago: University of
Chicago Press, 2005).
3. Porter, Trust in Numbers (ref. 2).
4. Alain Desrosières, The Politics of Large Numbers: A History of Statistical Reasoning (Cam-
bridge, MA: Harvard University Press, 2002); Ian Hacking, The Taming of Chance (Cambridge:
Cambridge University Press, 1990).

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms
 DATA IN TIME | 583

became categories that shaped new perceptions of social, political, economic,

and scientific reality.
Two implicit corollaries of this turn toward aggregate statistical reasoning
stand out. The first is that the construction of broad statistical objects—
whether in economics or state administration or natural history—required
large amounts of data as grist for the statistical mill. The nineteenth century
was, as Hacking has put it, the era of an “avalanche of printed numbers,” and
armies of bureaucratically organized collectors funneled data on census figures,
weather observations, botanical and zoological specimens, and the like toward
centralized experts tasked with organizing and interpreting them.5 These pro-
jects were supported by new institutions (e.g., large statistical bureaus), new
technologies (e.g., index cards, paper slips, and eventually punched cards), new
social arrangements (a hierarchical system in which many relatively untrained
“invisible workers” supplied raw material to fewer expert analysts), and new
disciplines (the state sciences, economics, the modern professionalized natural
sciences). As much as this process required trust in numbers, it also required
trust in data and in the patterns extracted from large data collections. We speak
now of the twenty-first century being the age of “Big Data,” but much of the
machinery—to say nothing of the underlying totalizing ambitions of data-
driven statistical projects—was established long ago. The nineteenth century
was an era of big data, too.6
The second is that in order for the results of statistical reasoning to become
legible and translatable to a relatively broad audience—and for statistical
methods to be standardized to the degree that they could be adapted across
many disparate disciplines—new methods of representing statistical arguments
were developed. Many of the characteristic elements of today’s statistical
language—particularly pictorial charts and graphs of all descriptions—were
invented or popularized during the later nineteenth century, an era one
historian has labeled “the Golden Age of statistical graphics.”7 The statistical
turn required a new visual culture that enabled people to “see” statistical
objects: graphical representations made visible entities that were composed
of abstracted data points or mathematical relationships. We take such

5. Ian Hacking, “Biopower and the Avalanche of Printed Numbers,” Humanities in Society 5
(1982): 279–95.
6. See, particularly, Christine von Oertzen, “Datafication and Visualization in Nineteenth-
Century Census Statistics,” in this Issue.
7. Michael Friendly, “The Golden Age of Statistical Graphics,” Statistical Science 23, no. 4
(2008): 502–35.

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms
584 | SEPKOSKI

visualizations to be intuitive today, but it took many decades over the nine-
teenth century for these conventions to be accepted, and their emergence was
tied to highly contingent convergences between practices in often disparate
fields of inquiry. How did this transformation come about?
For insight into this question, I will point to contemporary developments in
two seemingly disparate fields of knowledge: early nineteenth-century state
science (cameralism), and paleontology. As my central case study, I will high-
light one particular convergence: the emergence of visual approaches to patterns
in temporal data. My argument is that the temporalizing of data in graphical
form drew on developments in both traditions, but that the knowledge
exchange involved depended on actors whose interests were highly idiosyncratic
and divergent. This case study, then, offers both an insight into the contingent
circumstances around the establishment of a particular visual idiom, and an
advertisement for the usefulness of “history of data” as a category.
As Daniel Rosenberg and Anthony Grafton have shown in their innovative
study of the history of the “timeline,” the now-familiar convention that time
can be visually represented as a line passing through measured graphical space
did not fully emerge until the late eighteenth century.8 One of the most
influential early presentations of this visual idiom was Joseph Priestley’s
1764 A Description of a Chart of Biography, which depicted the lives of some
2,000 notable historical figures as horizontal lines measured against absolute
graphic space. As Priestly put it, “Thus the abstract idea of TIME . . . admits of
a natural and easy representation in our minds by the idea of a measurable
space, and particularly that of a line; which like time, may be extended in
length, without giving any idea of breadth or thickness.”9
This visual device caught on quickly and was often imitated, but in its initial
form it was somewhat limited: being two-dimensional, the only meaningful
information is conveyed on the horizontal (temporal) axis. However, others
quickly realized that the vertical axis could also be employed, enabling a graph
to demonstrate some change in a variable over time. Most notably, the Scottish
engineer and political economist William Playfair adapted Priestley’s conven-
tion to now-familiar line graphs, representing (for example) fluctuating
imports and exports in England. The importance of this innovation was that
it merged the geometric tradition of the Cartesian coordinate system

8. Daniel Rosenberg and Anthony Grafton, Cartographies of Time: A History of the Timeline
(Princeton, NJ: Princeton Architectural Press, 2010), 113.
9. Joseph Priestley, A Description of a Chart of Biography (London: J. Johnson, 1764), 5.

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms
 DATA IN TIME | 585

(developed in the seventeenth century) with Priestley’s projection of time as

graphical space.
Another important innovation of Playfair’s was that his graphs arranged
economic data in temporal series; that is to say, they presented a historical
account of a phenomenon composed of data points. But in his own day, this
epistemic convention did not catch on—at least in the economic sciences. At
the turn of the nineteenth century there was a well-established tradition in
Northern Europe and Britain of quantitative analysis of state resources, labeled
variously “cameralism” or “political arithmetic,” that attempted to quantify or
forecast state resources and population statistics using large quantities of data.
However, these approaches almost never utilized graphs, being instead focused
on the tabular presentation of data.10 Indeed, despite the apparently obvious
fact that “tables of statistics in general, and comparative political data in
particular, suggests the need for graphical comparison,” prior to the 1860s
graphical visualizations of statistics were not warmly welcomed into economic
arguments.11 When, for example, the English mathematical economist
William Stanley Jevons approached the noted statistician and publisher
William Newmarch in the early 1860s about publishing a “Statistical Atlas”
he was composing, he was taken aback by Newmarch’s profound disinterest in
his statistical visualizations. As economic historian Haro Maas puts it,
“Apparently, it was not at all obvious for someone like Newmarch to present
a table of numbers in a graph.”12
However, one place where narrative graphical summaries of data were fairly
well-established prior to the mid-nineteenth century was taxonomic natural
history. For example, Alexander von Humboldt devised a number of innova-
tive visual methods for summarizing biogeographical and climactic informa-
tion, such as the image depicting the geographical distribution of plants on the
island of Tenerife in Figure 1. Although somewhat stylized, this image is,
effectively, a line graph: as von Humboldt himself acknowledged, “This graph-
ical method is analogous to what M. Playfair first employed,” noting that “the
map which M. Playfair gives of the national debt of England brings to mind
the section of the Pic of Teneriffe.” Graphical techniques such as these were
essential to Humboldt’s communication of complex relationships in data

10. James R. Beniger and Dorothy L. Robyn, “Quantitative Graphics in Statistics: A Brief
History,” The American Statistician 32 (1978): 2.
11. Ibid., 3.
12. Harro Maas, William Stanley Jevons and the Making of Modern Economics (Cambridge:
Cambridge University Press, 2005), 218.

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms
 586 | SEPKOSKI

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms
FIGURE 1: Alexander von Humboldt’s “line graph” showing distribution of plants on the island of Tenerife. Source: Alexander von Humboldt, Viage á las
regiones equinocciales del nuevo continente: Hecho en 1799 hasta 1804, por Al. de Humboldt y A. Bonpland (Paris: En Casa de Rosa, 1826), vol. 1;
reproduced in Humboldt, Political Essay on the Kingdom of New Spain (London: Longman, 1814), vol. 1.
 DATA IN TIME | 587

since, as he put it, “statistical projections that speak to the senses without
fatiguing the mind, possess the advantage of fixing the attention on a great
number of important facts.”13
Geologists, too, had a fairly established visual culture by the early decades of
the nineteenth century. Whereas some of the most famous of these images
were spatial representations of the stratigraphic layers of the earth, arranged
either as topographical maps (William Smith) or idealized stratigraphic
columns (Georges Cuvier and Alexandre Brongniart), others followed the
tradition of Priestley and Playfair by representing temporal data in graphical
space.14 One of the first true line graphs in paleontology was published in 1860
by the English geologist John Phillips (who, interestingly, was the nephew and
collaborator of the stratigraphical pioneer William Smith), depicting changes
in the taxonomic diversity of fossils over the history of life (Fig. 2). Phillips’s
image abstracted a pattern of the waxing and waning of life’s diversity from
a compendium of taxonomic data in a manner precisely analogous to
the method that had been employed by Playfair and by contemporaries like
William Jevons—it revealed a complex temporal phenomenon in data that
would otherwise have been invisible.15
Whereas Humboldt was quite open about the source of his graphical inspi-
ration, the story in geology and paleontology is a bit more complex. Phillips’s
graph was, in 1860, published at a moment when similar graphs were becoming
more popular in political economy, demography, and the physical sciences,
and Phillips might have had any number of models to draw on. But similar
kinds of narrative visual summaries had already emerged significantly earlier, in
the late 1830s and 1840s, well before such visualizations of economic statistics
had become broadly accepted. One particularly significant example was the
work of the German naturalist Georg Heinrich Bronn, a paleontologist who
occupied a chair at Heidelberg and was considered in his day to be one of the
leading scientists in Europe. Bronn advocated a distinctive approach to the
study of life that was explicitly statistical in nature: basically, he spent his career
tabulating data on all known taxonomic groups of fossils, which he compiled
in huge compendia, like his 1849 Index palaeontologicus: Oder Übersicht der bis

13. Alexander von Humboldt, Political Essay on the Kingdom of New Spain, vol. 1 (London:
Longman, 1814), cxxxii–iii.
14. Georges Cuvier and Alexandre Brongniart, Description géologique des environs de Paris
(Paris, 1812); William Smith, Strata Identified by Organized Fossils (London,: W. Arding, 1816).
15. John Phillips, Life on the Earth (London: John Murray, 1860).

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms
588 | SEPKOSKI

FIGURE 2: John Phillips’ depiction of the history of life as a line

graph tracing three successive curves. Source: John Phillips, Life
on the Earth (1860), 66.

jetzt bekannten fossilen Organismen (Index of Paleontology: or an Overview of

the Hitherto Known Fossil Organisms).16
Bronn’s method was to convert standard taxonomic entries found in pale-
ontological catalogs and monographs into numerical data, which he then
subjected to a variety of statistical analyses (e.g., calculating average durations
of species, ratios of species to genera at different periods, relative composition
and distribution of these groups in time). These relationships were expressed,
in the first instance, as numerical tables, much like the tables favored by the
cameralists and political arithmeticians of the late eighteenth century. There

16. H. G. Bronn, Index Paleontolgicus, oder Übersicht der bist jetzt bekannten Fossilen Orga-
nismen (Stuttgart: 1849).

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms
 DATA IN TIME | 589

was a very good reason for this approach: Bronn, though ultimately appointed
to the first Chair in Zoology at his university, was in fact trained not as
a naturalist, but as a practitioner of the bureaucratic science of Kameralwis-
senschaft (cameralism). Indeed, Bronn received his habilitation in that subject
from the University of Heidelberg in 1821, and immediately joined the faculty
in the school of cameral studies. He did not publish on geological topics until
the early 1830s, although his particular focus in state administration was
forestry and agriculture, which led him to travel widely, observing natural
environments in Germany, Swizerland, and Italy that ultimately seems to have
inspired his interest in fossils.
However, it is noteworthy not just that Bronn was a cameralist, but espe-
cially that he was trained and taught at Heidelberg during a period of transition
in economic state sciences. Although it has been fairly extensively documented
that most cameralist practice—and, indeed, statistical practice at the time more
generally—viewed the accumulation of numerical data as a purely descriptive
enterprise that explicitly avoided seeking explanatory patterns and laws, as
Marco Tamborini and I have shown elsewhere, the Heidelberg school was
an exception.17 Specifically, Heidelberg cameralists followed the German
agronomist Albrech Daniel Thaer and his student Carl von Wulffen in dis-
tinguishing between “statistics” and “statics,” the latter of which they defined
as an approach explicitly contrasted to what was perceived as the dry collection
of data for data’s sake. As practiced at Heidelberg—for example, by the forestry
expert Johann Christian Hundeshagen—cameralism involved a two-part pro-
cess: first the data was collected and tabulated, which was statistics, and next
mathematical relationships within the tables were calculated, which was statics.
These lessons were absorbed deeply by Bronn. His compendia of fossil data
were assembled not just to document the empirical record of fossils, but to
determine relationships among the data that could reveal patterns in the
history of life, and Bronn seems to have been aware of the novelty of his
approach. In his first paleontological study—an analysis of the succession of
strata in the Italian Tertiary formations—he concluded that “these studies are
sufficient not only to settle a dispute concerning the Italian Tertiary structure,
but also to demonstrate the application of a numerical approach to character-
istics of the fossil deposits in rock strata, that has so far not been considered.”18

17. Sepkoski and Tamborini, “‘An Image of Science’” (ref. 2).

18. Heinrich Georg Bronn, Italiens Tertiär-Gebilde Und Deren Organische Einschlüsse: Vier
Abhandlungen (Heidelberg: Groos, 1831), 175 and 74.

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms
590 | SEPKOSKI

This was, for Bronn, the beginning of a new discipline, which he appropriately
termed “paleontological statics.” The need for better data prompted him to
devote his career to assembling his massive taxonomic data compilation,
which he believed would provide a firm basis for analyzing the patterns and
regularities in the history of life. Whereas a number of other naturalists of
Bronn’s day—in disciplines ranging from botany to zoology to paleontol-
ogy—had compiled numerical tabular accounts of species and genera, Bronn’s
approach was distinctive because it applied statistical analysis to a causal expla-
nation of the data he collected—and this was the influence of his cameralist
training in statics.
But Heidelberg cameralists generally did not employ pictorial summaries of
data in the style of Playfair or von Humboldt—their visual culture was still
firmly rooted in the tabular idiom. In his more mature paleontological stud-
ies—the 1849 Index, for example—Bronn introduced visualizations of his data
patterns, like the one reproduced in Figure 3. This image is one of the earliest
examples of what is now known as a “spindle” diagram, one of the most
common visual devices in paleontology for depicting the duration and diver-
sity of a higher taxon as a line of varying thickness (according to the number of
constituent taxa at a given time). This particular diagram depicts the history of
major groups of organisms throughout geological history, and effectively takes
the form of a visual historical summary. We see, at a glance, the rise and
diversification of the ammonites before their abrupt departure alongside the
waxing and waning of the brachiopods, vascular plants, crustaceans, mammals,
etc. The picture, then, tells a story about the history of life, and has become
a standard tool in evolutionary biology—and especially paleontology—to
this day.
What is significant about this image is not just that it reflects the incorpo-
ration of statistical visual practices to natural history, but that it also signals an
important epistemic shift in thinking about—and visualizing—history itself.
In describing the somewhat later innovations of Jevons in economic statistics,
Maas observes that “before graphs could reveal economic phenomena and
feature in economic explanations. . . . [h]istorical events had to be repackaged
as data,” something Maas labels “the timing of history.”19 This is precisely
what Bronn, drawing on his cameralist training, saw was necessary for narrat-
ing broad patterns in the history of life. In the first place, historical “events”
(i.e., the individual life histories of organisms) were aggregated and converted

19. Maas, William Stanley Jevons (ref. 12), 220.

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms
 DATA IN TIME | 591

FIGURE 3: One of Bronn’s spindle diagrams depicting the relative changes in diversity
among different groups of animals throughout geological time. Source: H. G. Bronn, Index
Paleontolgicus, oder Übersicht der bist jetzt bekannten Fossilen Organismen (Stuttgart:
1849), 778.

to data points representing taxonomic units that could be counted and treated
statistically. Secondly, those data were projected as a temporal progression in
visual space, “stabilized” not only along the horizontal (temporal) axis, but also
on the vertical, which represented change in quantity. Indeed, it is only
through the abstraction of individual events to commensurable data points,
and the aggregation of those data, that it is possible to identify the regularities
or “empirical laws” that Bronn hoped to discover—in this case, what Bronn
would later describe as the tendency for major taxonomic groups to progres-
sively diversify, reach a maximum point, and then slowly decline, only to be
succeeded by similar cycles in other groups.20

20. H. G. Bronn, Untersuchungen Über Die Entwickelungs-Gesetze der organischen Welt Wäh-
rend der Bildungs-Zeit unserer Erd-Oberfläche (Stuttgart: Schweizerbart’sche Verlagsbuchhan-
dlung, 1858), 489. Interestingly, the historiographic conventions implicit Bronn’s statistical
visualizations also resonate suggestively with contemporary philosophies of human history, such
as works by G. F. Hegel and the English historian Henry Buckle, the latter of whom promoted
a “statistical” approach to history in his History of Civilization in England (London: John Parker,
1857). Notably, Buckle’s work received a great deal of attention in Germany, where it was read
appreciatively by statisticians but often criticized for, as Prussian Statistical Bureau director Ernst
Engel asserted, confusing statistical regularity with law. Engel’s view—which is discussed in detail
by Christine von Oertzen in this Issue—reflected an important contemporary debate about the

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms
592 | SEPKOSKI

My purpose in highlighting Bronn’s statistical visualization of natural his-

tory in the context of contemporary economic statistics is to give an example of
what I believe “history of data” is most useful for: that is, illuminating the
complex web of practices, influences, and contexts associated with the devel-
opment of modern information cultures. I have not attempted to tell a heroic
story of discovery here, but rather to suggest that behind a humble graph—
Bronn’s spindle diagram—is a complex story involving a series of important
transformations in the way Western science has approached issues as disparate
as quantification, history, visual language, and statistical abstraction. The
establishment of conventions around the legitimacy of statistical inferences
and the visual representation of aggregative temporal patterns has had a pro-
found effect on the shaping of modern politics and culture, as well as science.
Though few historians today would attempt a statistical explanation for major
trends in human history, in the historical sciences—including the economic
and social as well as the natural sciences—statistical language is pervasive.
Although relatively little attention has been paid to the implicit historiography
of those disciplines, it strikes me that a catholic approach to the history of data
is an ideal means of addressing it more closely.
Another important lesson is that visualizations of quantitative data invoke
the power of numbers without requiring advanced numeracy in the viewer. As
Porter and others have pointed out, trust in numbers became a kind of implicit
trust, and in visualizations the numerical data and their analysis are effectively
black-boxed. Statistical visualizations make patterns in data visible, but they
also obscure the phenomena that stand behind the data themselves. Bronn’s
spindle diagram may have made it easier for his reader to grasp the statistical
narrative he hoped to convey, but unlike his data tables or fossil catalogs, the
story the images tell is unequivocal. In the sciences, this has recently become
a source of tension, where in a variety of disciplines stark divisions—often
associated with status, credit, and access to funding—have emerged between

-
viability of statistics for historical reconstruction in natural history and other disciplines as well. In
the decades following Bronn’s death (in 1862), German-speaking paleontologists debated whether
an aggregative statistical approach was valid for identifying historical patterns in the fossil record.
Like Engel, paleontologists Melchior Neumayr and Karl Alfred von Zittel argued that statistical
analysis could not identify causal relationships that are the proper foundation for establishing laws
of nature, and urged instead a focus on reconstructing individual “snapshots” of life’s history
through careful study of individual fossil groups. See Hacking, Taming of Chance (ref. 4), 125–29;
Marco Tamborini, “Paleontology and Darwin’s Theory of Evolution: The Subversive Role of
Statistics at the End of the 19th Century,” Journal of the History of Biology 48 (2015): 575–612.

This content downloaded from

189.147.46.136 ff:ffff:ffff on Thu, 01 Jan 1976 12:34:56 UTC
All use subject to https://about.jstor.org/terms
 DATA IN TIME | 593

data “producers” (i.e., bench scientists) with deep understanding of phenom-

ena and “analysts” (or “data jockeys”) who are sometimes accused of glib
disinterest in anything other than correlations and models.21 In political and
economic contexts the disconnect between data and visualization or aggregate
summary is potentially more striking and perilous: in some examples of truly
huge data (such as Google analytics of web activity) the data are so vast and so
widely distributed that no person can claim to really know it. Trust in this
kind of data amounts to trusting the algorithms that handle it for us.
A genealogical approach to data history can help us better understand, if not
escape, the implicit choices and compromises that have been made in seeing
the world as data.

ACKNOWLEDGEMENTS

I would like to thank Soraya de Chadarevian and Ted Porter for the invitation to
contribute to this special issue and the workshop that preceded it. Thanks also to
the members of the initial workshop for thoughtful comments about my oral presen-
tation, and to Soraya, Ted, Hallam Stevens and Christine von Oertzen for helpful
comments on the written draft. Portions of this article have drawn on research con-
ducted in collaboration with Marco Tamborini; I would like to acknowledge his impor-
tant role in the larger project on which this essay is based.

21. See, for example, Bruno J. Strasser, “Data-Driven Sciences: From Wonder Cabinets to
Electronic Databases,” Studies in History and Philosophy of Biological and Biomedical Sciences 43
(2012): 85–87.

This content downloaded from

189.147.46.136 on Wed, 07 Sep 2022 05:34:47 UTC
All use subject to https://about.jstor.org/terms