












































[Expositions 6.2 (2012) 15-22]  Expositions (online) ISSN: 1747-5376 

Getting “Wrong” Right  
 
COURTNEY ROBY 
Cornell University 

 
Daryn Lehoux begins What did the Romans Know? An Inquiry into Science and Worldmaking 
(2012) with the broadest of questions: “What’s in a world?” In the chapters that follow he 
examines the spectrum of answers constructed by a selection of Roman authors, analyzing how 
the differences between their answers (and between theirs and ours) might be predicated on 
different ways of seeing the world in front of us and different systems of “facts” constructed 
from what we see. He devotes particular attention to Roman accounts of the natural world which 
have typically been ignored or quickly dismissed as “wrong” (including astrology, divination, 
and the devastating effects of garlic on magnets), making these “errors” yield up broader 
epistemological truths.  

The reader is immediately warned against taking for granted too much uniformity between 
“scientific” cultures, and particularly against assuming a one-to-one correspondence between an 
observer’s truth and things that are actually present in the world. This search for difference, for 
Lehoux, lies at the heart of the enterprise of history of science: “we could have done it 
differently. Indeed, on one way of looking at it, the history of the sciences is virtually a catalogue 
of different ways of doing it…” (Lehoux 2012, 1) Estrangement of Roman thought, and of 
ancient science more generally, is a crucial part of Lehoux’s strategy for allowing his reader to 
see Roman scientific thought through new eyes. 

Each chapter is structured around a particular problem and an exemplary text or two: Seneca’s 
Natural Questions informs a discussion of the extent to which a judicial system which relied on 
rhetoric influenced the exploration of nature in the early Empire; Ptolemy’s Tetrabiblos and 
Manilius’s Astronomica yield a spectrum of possible models underlying Roman beliefs in 
astrology; and so on. The final two chapters are completely dedicated to the problems a “realist” 
standpoint creates for the history of science, and to alternative theories of truth and their 
explanatory power for both modern problems in scientific epistemology and the now-familiar 
Roman problems explored elsewhere in the book. Where possible, Lehoux links the discourse in 
the ancient texts to persistent puzzles in epistemology and philosophy of science: the importance 
of witnessing, the validity of analogical models, the requirements for “laws” in nature, and so 
forth.  

What Did the Romans Know? is not exactly a book about the history of Roman science, not 
even if this history is defined broadly in sociological and material terms as well as those of more 
traditional intellectual history. It is rather about the philosophical, linguistic, and rhetorical terms 
of the discourse of epistemology as it evolved around a series of particular problems. The book is 
grounded not so much in Roman social realia or scientific or textual practices, as in a set of 
philosophical “worldviews” which differ more or less radically from our own, though they 
operate on familiar objects (which Lehoux exemplifies at the book’s end as “walls and eyeballs, 



Getting “Wrong” Right  16 
 

stars, nerves, paintings, and perfume” (Lehoux 2012, 232)). In this sense it is a very different 
work from, for example, Serafina Cuomo’s Technology and Culture in Greek and Roman 
Antiquity (2007), though the two books bear a superficial resemblance in their organization into 
chapters treating a series of disparate scientific or technological topics from the ancient world 
using a series of disparate analytical frameworks.  

Multiple-parameter variation of this kind allows more ground to be covered than could ever 
be managed through a more methodical approach. The danger, of course, is that the reader will 
fill in the blanks incorrectly, left without guidance as to how applicable the historical or 
philosophical conclusions of one chapter are to the different social or intellectual contexts of the 
others. Lehoux’s treatment sidesteps this danger for the most part through cross-references in the 
chapters that establish continuity between their various problems and methodologies. For 
example, he returns several times to examine why the curious belief that garlic inhibits the power 
of magnets turned out to have such longevity. The later chapters also devote progressively more 
space to methodological and broad-based epistemological considerations, picking up threads 
from earlier chapters to illustrate the breadth of conclusions where possible.  

Particularly eloquent on these matters is the fifth chapter, on “The Embeddedness of Seeing.” 
In this chapter Lehoux addresses the familiar and deceptively tricky mirror-reversal problem: 
why does a mirror reverse left and right, but not up and down? The chapter’s aim is not to 
provide an explanation for this problem (though Lehoux does this as well), but to address 
something more interesting: the apparent behavior of objects in a mirror simply does not 
comprise a problem for Ptolemy in the same way as it does for us. Lehoux accounts for this by 
saying that the mirror-reversal “paradox” can be viewed as a product of a modern division of the 
study of sight into its physical-optical, physiological, and psychological components. Ptolemy’s 
optics was a different kind of science, in which these components were more unified, and as a 
result “Ptolemy does not have the same blind spots as I do” (Lehoux 2012, 108).  

Discussion of Ptolemy’s account of seeing (and later of Galen’s) gives way to a broader 
problem on different “epistemologies of seeing” (Lehoux 2012, 125). Lehoux observes claims by 
Skeptical philosophers, especially Pyrrhonians, on the dubious nature of many seen things, 
particularly given the idea that different people (of different constitutions, etc.) may see things 
differently, just as humans and dogs may. In the course of this discussion the elegant efficiency 
of the book’s argument becomes particularly clear. This problem mirrors the challenge of the 
book as a whole: we and the Romans may see things differently, so we ought not to read our own 
experiences too much into these debates, striving instead to understand these debates on their 
own terms. While each of the problems starring in the individual chapters drive this point home 
to some extent, here the value of the exemplary ancient problems in motivating a coherent, 
generalized epistemological investigation becomes especially plain.  

In particular, Lehoux argues that the view that “people believed things – nay, even saw things 
– because they were true, because they were really out there in the world to be seen” (Lehoux 
2012, 14) is itself problematic. Taxonomy is very important to Lehoux’s answer to what turns 
out to be the book’s central question (which he draws from Bas van Fraassen): “Is there any 



17  Roby 
 

rational way I could come to entertain, seriously, the belief that things are some way that I now 
classify as absurd?” (in Lehoux 2012, 152; van Fraassen 2002, 73). The shift in plausibility of 
the garlic-magnet hypothesis, for example, is fundamentally predicated on a shift in 
categorization: “magnets used to be the kind of thing that was sympathetic, as was garlic. Now 
magnets are the kind of thing that are magnetic, and garlic in our experience is not” (Lehoux 
2012, 151). These categories are not simply labeled drawers for the things the world is made of; 
taxonomy brings with it a whole theoretical infrastructure explaining how the things in a given 
category are related to each other.  

This infrastructure, in turn, is generally the product of induction, and this has typically been 
enough reason for modern philosophers of science committed to the “realism” of laws and 
theories to reject these approaches to understanding the world as quaintly wrong-headed. In turn, 
this frequently results in the disqualification of ancient investigations into nature from 
membership in the class of “scientific” practices in the proper sense of the word (on this, see the 
excellent discussion of “laws of nature” in the third chapter). The alternative has traditionally 
been a cultural relativism which makes room for such beliefs at the cost of intelligibility: 
scrupulously avoiding interpreting Roman “scientific” thought according to the rules of modern 
scientific practice is good as far as it goes, but what means does it give us for understanding the 
practices of a culture so remote from our own?  

Lehoux takes a third path, which he labels “coherence”: he asks not how well Roman science 
conforms to science as it happens to be practiced generally today, but how well Roman 
descriptions of objects in the world and the relationships between them cohere with one another. 
The criterion of truth that goes along with this approach is that “we count as true any 
propositions that have passed the best tests we can think of to throw at them” (Lehoux 2012, 
243); Lehoux identifies this as a “pragmatic” theory of truth. Pragmatic it certainly is: the tests of 
truth can (and do) change with time and practitioners, and the focus on what “we believe” is true 
allows for a greater breadth of discourse than the realist “what is true.” Is the pragmatic theory 
satisfying, as a structure that might be generally applied to ancient scientific practices other than 
those described here? Lehoux makes a very good account of it, and of the need to “read” theories 
into the world (Lehoux 2012, 239) rather than reading them “back” to axiomatic foundational 
claims. At the same time, the scope of the book quite reasonably does not allow for a full 
discussion of the justifications for such a theory, which is always at risk of slipping back toward 
relativism if the criteria for testing and believing are not moored quite tightly. 

In rejecting the received wisdom of a 20th-century philosophy of science which strives at all 
costs to fit scientific theories into a deductive model, Lehoux’s approach can be constructively 
juxtaposed with that of Rom Harré’s The Principles of Scientific Thinking (1970). This work 
critiques the “deductivist” view of philosophy of science, in which logical order is taken to 
match up with the natural order of phenomena being studied, and mathematics to be an 
approachable ideal for scientific knowledge of all kinds. Harré identifies several problems with 
this approach, beginning with the assertion that modern scientific theories truly have the degree 
of deductive rigor attributed to them by those who subscribe to the deductivist view. He begins 



Getting “Wrong” Right  18 
 

his critique with a question strikingly similar to Lehoux’s motivating query: “Of what nature is 
the world scientists attempt to describe and understand?” (Harré 1970, 10). While Harré would 
later be associated with a kind of scientific realism, and might therefore appear at first glance to 
be at odds with Lehoux, he went on to distinguish a variety of types of realism, arguing in an 
essay in Derksen’s The Scientific Realism of Rom Harré (1994) for a realism that depends on 
inductive arguments working over type-hierarchies in order to create models. The seeds of this 
version of realism are already present in The Principles of Scientific Thinking (1970), and it is 
those aspects which can be most usefully compared with Lehoux’s approach. 

Theories tell us why patterns of phenomena are the way they are. In order to do so, a theory 
often has to provide us knowledge we do not previously have; it does so by providing a model 
for the unknown. This model is patterned after objects and processes that are already known, and 
at a rough approximation is hypothesized to work as an analogue for the real, unknown, 
mechanism. The model is absolutely central to Harré’s epistemology, while a system of 
deductive laws is what he refers to as “a desirable heuristic device” (Harré 1970, 2). This is an 
inversion of the “deductivist” view, in which models are at best secondary to a system of formal 
logic which is taken to reflect the rules governing the phenomena themselves. In practice, Harré 
argues, constructing such a model requires tools other than the logic of statements: “If this logic 
is used as the exclusive tool of analysis, it imposes an unnatural structure upon theory” (Harré 
1970, 42). The sentential explanans cannot stand on its own, as the deductivist view suggests: it 
requires additional information in order to refer to new objects (or processes) rather than familiar 
givens, and so analogy is required to make the statement meaningful.  

Harré’s rich descriptions of these analogical processes provides useful background to 
Lehoux’s appeal to building categories based on perceived likenesses as a tool for generating 
theories about the natural world, and in particular the questions Lehoux raises about the degree 
of difference between ancient and modern scientific practices and standards. Taxonomy is 
extremely important to the criteria of truth Harré develops, particularly in his articulation of the 
“transforms” that link the expression of a model with that of the phenomena it represents. For 
example, Harré describes the discovery that salt crystals are lattices of sodium and chlorine ions 
as a “modal transform” between the shape of the crystal and its internal lattice structure, a 
description likely to prompt classification of salt crystals with other crystals of a similar lattice 
structure. He contrasts this description of salt, labeled as “sodium chloride,” with one that might 
arise in a domestic context, where it might be classified alongside peppercorns, bay leaves, and 
parsley. Depending on the disciplinary context, Harré argues, "the taxonomic weight cannot so 
easily or so plausibly be placed upon the technically most advanced concepts" (Harré 1970, 54). 
Even in the practice of modern science, then, taxonomy is not a simple matter. Harré opposes the 
“positivist myth of infinite arbitrariness of classification” to the “realist” stance (further specified 
in Harré 1994 as “entity realism”) in which things and substances have real differences in their 
constitutions which are manifested in external characteristics, and works to find ground between 
these two extreme stances which seems to correspond to the realities of scientific practice and 
discourse.  



19  Roby 
 

Ultimately Harré subsumes the criteria of truth that fall under the categories of 
correspondence, coherence, and pragmatic success under a single larger criterion which he 
argues is met by all of them. This overarching criterion requires that the sentence used to make a 
true statement “is constructed in such a way and of such elements as are conventionally used to 
locate and individuate something where the subject of the statement actually is, and to ascribe to 
it a property which it actually has” (Harré 1970, 189). It is not my purpose here to address the 
validity of this claim with regard to Lehoux’s definition of “coherence” and “pragmatic truth,” 
but rather to note that the interested reader might use Harré as a guide in order to put them in a 
more detailed context. Harré later “type-based inductive argument for scientific realism” (Harré 
1994, 8) invokes a form of “realism” which explicitly recognizes the model-building value of 
observations made within a certain hierarchy of types. 

Harré also challenges the view that the “vehicles of thought” must necessarily be sentential 
statements, or indeed any kind of linguistic structures, and the concomitant belief that statements 
expressed this way have a particular logical validity in explaining phenomena in the world. He 
proposes instead that scientists in fact work with a “complex” of vehicles of thought which 
include linguistic formulations as well as visualizations, in all cases focusing on these vehicles as 
thought processes rather than reflections of outward phenomena, and on their outward 
manifestations, such as pictures or speech (uttered or written), as manifestations of internal 
thought processes and not of the phenomena themselves. The model, again, is central here: the 
model stands in for the natural mechanism (known or unknown); it may be expressed in logically 
linked sentences, but those sentences are themselves neither the model nor a direct reflection of 
the rules underlying the phenomena the model describes. The flexibility Harré’s system allows in 
defining the tools of scientific thought harmonizes productively with Lehoux’s call for a broader 
range of patterns of thought and speech to be classified as tools seen by their practitioners as 
appropriate for investigation of the natural world.  

The complementarity of these two approaches is made particularly clear in their treatments of 
“laws of nature.” Lehoux’s approach rejects the stipulation that using a word that can be 
translated into English as “law” to refer to the mechanism behind some natural phenomenon is 
either necessary or sufficient for “laws of nature” in a particular worldview to be acknowledged. 
Instead, the criterion ought to be that “lawlike statements” (Lehoux 2012, 74) are made about a 
given natural phenomenon. Lehoux quite rightly calls attention to the importance and difficulty 
of defining “lawlike” in this context. Ultimately, the examples that he cites as successful in this 
vein are “intelligible, measurable, and predictable” (Lehoux 2012, 70), thus far matching the 
requirements stipulated by Jane Ruby (Ruby 1986, 350).  

Lehoux suggest that a “lawlike” statement may further be portrayed as operating on 
mathematically-framed objects (e.g. “circular epicyclic and circular deferential” (Lehoux 2012, 
72)) and generating mathematical relationships (e.g. the apparent planetary station that occurs at 
a given point), without incurring any objection from Lehoux that it interferes with its explanatory 
power for real objects in the natural world. However, his conception of the “lawlike” also allows 
the mechanism to be divine, thus permitting discussion of the “laws” perceived to govern 



Getting “Wrong” Right  20 
 

practices like astrology and divination. Lehoux makes a strong argument for linking discourse 
about these practices with that about practices which correspond more closely to those currently 
classified as scientific. This argument is predicated at least in part on the acknowledgement that 
even among the disciplines now uncontroversially counted as sciences there are significant 
differences in what it means to say phenomena obey “laws of nature.” Lehoux uses biology and 
physics to stake out this spectrum, motivating this problematization with the famous dismissal 
(possibly Rutherford’s) of sciences other than physics as “stamp collecting” (Lehoux 2012, 75).  

These disagreements suggest more than anything else that there is work to be done on the 
modern side as well in defining what “laws of nature” mean for different scientific practices, and 
here too Harré’s analysis may prove useful. He emphasizes the importance of understanding the 
mechanism behind a process in valorizing induction-based “laws of nature” behind it. Laws, in 
this view, are causal laws, defined by their “universality,” “necessity” (which emerges from the 
generative mechanism), and “assent” (that is, generative power must be justified by reference to 
a generative mechanism). We recognize patterns in nature (like cubical salt crystals) and explain 
them in terms of what Harré calls the “powers” of the objects involved (like those of the sodium 
and chloride ions which make them bond in certain shapes). “Laws of nature” are then the 
statements that describe these patterns; at some level they involve the “generative mechanisms” 
(Harré 1970, 125) which result in phenomena in a particular state. Such a law (which may or 
may not be strictly verbal) must provide an account of a causal mechanism, or be demoted to a 
“protolaw.” These criteria, like those Lehoux proposes, are more inclusive than restrictive, and 
hence allow for discourse both about the exact sciences and other possibly scientific activities 
(e.g., evolutionary biology, psychology, astrology). Harré makes no requirement of mathematical 
expression, or even of exclusively symbolic representation; the focus is on accounting for some 
causal mechanism, whose nature is not precisely stipulated.  

Like Harré, Lehoux focuses on continuities of content rather than terminological coincidence. 
His approach to the question of whether “laws of nature” predate the early modern period, for 
example, rejects as invalid the grounds for Ruby’s assertion that nothing equivalent to a “law of 
nature” existed for medieval Islamic science. Ruby bases this claim on the fact that “Arabists tell 
me there is no word corresponding to ‘law’ in Arabic” (Ruby 1986, 344 n. 19). Lehoux 
eloquently refutes this by arguing that “If the history of the Frisbee cannot be pushed any farther 
back than the changing of its name from Pluto Platter, then we are talking about the history of a 
word, not of a thing” (Lehoux 2012, 68 n. 57). Neither Lehoux nor Harré denies the importance 
of linguistic structures for formalizing and propagating scientific knowledge, but both are 
insistent on the need to drill beneath the surface of such structures in search of the conceptual 
frameworks which underlie them. Both authors find problematic the requirement that in order to 
qualify as “science,” investigations into phenomena must be expressible as sentences which 
conform to deductive logic. Lehoux objects because this view disqualifies a great deal of ancient 
investigation into nature which seems otherwise to map fairly well onto activities we usually 
think of as “scientific,” while Harré argues that if one is scrupulously attentive to scientific 
practice, it disqualifies a fair amount of what is labeled “science” today as well.  



21  Roby 
 

Both views likewise call into question what is meant by a scientific explanation. Lehoux’s 
approach is modeled on Philip Kitcher’s definition of an explanation as a member of a set of 
sentences which “collectively provide the best systematization of our beliefs” (Lehoux 2012, 
180; Kitcher and Salmon 1989, 430). Harré (1994) attaches a comparable measure of 
“ontological plausibility” to the success of a process of induction over theories or models, only 
then judging a theory’s degree of empirical success. Harré (1970) claims in a more detailed 
argument that an explanation is just part of the way a theory explains phenomena; the theory is 
based on a model and consists of sentences linked to the model by “transforms,” as in the case of 
sodium chloride above. He emphasizes in particular that the deductivist view claims a symmetry 
between explanation and prediction which simply is not there. In neither case is a successful 
explanation defined as the end product of a chain of deductive reasoning performed on 
sentences, which happens to mirror a causal system present in the phenomena under 
investigation. Indeed, both authors problematize the equation of predictive success and 
explanatory power – Harré on the grounds that there are cases in modern or recent science in 
which one but not both is found, Lehoux on the grounds that any theory succeeds at prediction 
until it doesn’t, and that the predictive successes of modern science are measured on the order of 
decades at best, which looks rather paltry by the standards of ancient science.  

The eclectic approach of Lehoux’s project is justified, at least in part, by his assertion that 
“the Romans were of course not looking for a comprehensive theory of explanation itself – that 
question is very much a product of twentieth-century philosophical concerns” (Lehoux 2012, 
180). Lehoux does not insist that the diverse scientific problems and intellectual contexts he 
invokes throughout the book add up to a comprehensive and uniform entity called “Roman 
science,” whose players all follow the same rules – seeking to discover and describe those rules 
would be a quixotic enterprise, to say the least. Instead, he provides a series of well-chosen 
examples – the very stuff of which philosophy is made, as he observes (Lehoux 2012, 87) – and 
uses them to carve out a space dedicated to the consideration of complex questions about the 
complex enterprise of scientific thinking in the Roman world.  

The resulting work is not (and does not claim to be) the last word on the questions it raises; it 
is rather an incitement to further work on them and the new questions they raise in turn. This 
epistemologically sophisticated interrogation of Roman “scientific” activities represents an 
exciting opportunity for a new beginning in the dialogue between philosophy of science and the 
history of scientific practices in the ancient world. Works like Harré’s, which foreground the role 
of taxonomy in scientific model-building practices, provide some useful auxiliary tools for 
investigating the fascinating question of what “wrong” can tell us. 

 
 

Works Cited 
 

Cuomo, Serafina. 2007. Technology and Culture in Greek and Roman Antiquity. Cambridge, 
UK: Cambridge University Press. 



Getting “Wrong” Right  22 
 

Harré, Rom. 1970. The Principles of Scientific Thinking. Chicago: University of Chicago 
Press. 

–––. 1994. “Three Varieties of Realism.” In The Scientific Realism of Rom Harré. Ed. 
Derksen, Anthony A. Tilburg, The Netherlands: Tilburg University Press. 

Kitcher, Philip, and Wesley C. Salmon. 1989. Scientific Explanation. Minneapolis: 
University of Minnesota Press. 

Lehoux, Daryn. 2012. What Did the Romans Know?: An Inquiry into Science and 
Worldmaking. Chicago: University of Chicago Press.  

Ruby, Jane E. 1986. “The Origins of Scientific “law”.” Journal of the History of Ideas. 47.3: 
341-359. 

Van Fraassen, Bas. C. 2002. The Empirical Stance. New Haven, CT: Yale University Press, 
2002. 


