Science Outside the Laboratory, by Marcel Boumans, presents a fascinating journey through various episodes of economics and other sciences, mostly from the twentieth century. The author discusses these episodes in relation to problems of measurement and expertise. I have read the book several times, on planes and trains, on beaches and balconies, in the office and at home: with every read, the depth and detail of the historical accounts informed and educated me in various, and often unexpected, ways. For its accomplished scholarship in the history of science alone, the book belongs on the bookshelf of any philosopher of science who is interested in the scientific practices of the social sciences. Boumans’s stated goal is to advance what he calls “an account of measurement for the field sciences” (24), which is supposed to cover sciences that perform observation and measurement “outside the laboratory.” This account I found harder to come to terms with. For one, it remained by and large implicit in the specific episodes Boumans discusses. For another, those elements of the account that were explicitly stated can be questioned from a philosophy of science perspective. In what follows, I charter my attempts to grasp the account of measurement for field science that Boumans aims to advance and raise some questions about it.

Boumans begins the book by first taking the reader through some of the main theoretical approaches to measurement. He introduces the main elements of the representational theory of measurement, which has famously characterized measurement as a mapping between empirical relations on the one hand and numerical relations on the other. On this account, we can speak of measurement whenever observable qualitative relations (e.g., comparing the length of objects) obey a number of conditions such that they can be expressed numerically (e.g., by saying that the length of an object is 15 inches). Boumans argues that this account of measurement is too idealized to hold for measurement practices in the social sciences “outside the laboratory.” He proposes to analyze the specific methods and modeling strategies in the social sciences to arrive at a better understanding of how such sciences perform measurement. The remainder of the book discusses measurement practices in different areas in detail, such as meteorology, actuarial science, econometrics, and clinical judgment, among others. The discussion of these different cases is rich and insightful. Yet, it is less clear how they help to establish an account of measurement practices in the field sciences.

What the treatment of the different episodes does capture are what Boumans calls “methodological challenges of field sciences.” They boil down to this: field sciences conduct their observations and measurement in an environment, namely, the field, in which variables are not as easily controlled as they are within a laboratory. A methodology of field science thus has to specify how observation and measurement can be achieved in complex and sometimes uncontrollable conditions and has to consider problems of subjective judgment and interactions between scientists. While this seems an important perspective to advance, the distinction between science inside the laboratory and science in the field (or outside the laboratory) is not explained in great detail. We mainly learn about the distinction indirectly or implicitly. Indirectly, we can take all the cases that are treated in the book as telling us what “outside” the laboratory means. Implicitly, there are a few comments about the contrast between natural and social sciences that seem to be related to the distinction between “inside” and “outside” the laboratory. Indeed, in the conclusions of the book, Boumans relates the “methodological tension between laboratory and field science” (174) to W. E. Persons’ (“Some Fundamental Concepts of Statistics,” Journal of the American Statistical Association 19 [1924]: 1–8) equivocation between laboratory methods and the natural sciences on the one hand and field methods and the social sciences on the other. It is unclear whether Boumans actually endorses this equivocation. The evidence is mixed: he goes on to discuss other aspects of scientific method in a fashion that suggests his endorsement in the conclusion of the book, after having taken a much more careful stance in the beginning of the book. He also includes a few natural science cases as examples of field science, such as the meteorology case in chapter 3. At any rate, the distinction between “inside” and “outside” the laboratory is less clear than it would need to be in order to offer a way to understand the specific challenges of field sciences that Boumans claims to discuss.

In many ways, these ambiguities do not hinder the reader enjoying the masterful treatment of several interesting episodes in the history of economic thought, many of which are off the beaten path and thus interesting in their own right. They do, however, raise worries from a philosophy of science perspective. I will now focus on two of them.

The first worry relates to underdetermination and values. Specifically, the worry is that there is a lot of fruitful work in the philosophy of science that is related to underdetermination and values that has been originally conceived of as relating to empirical work (conducted inside the laboratory, in Boumans’s terminology). Boumans’s distinction between inside and outside the laboratory carries with it a real risk that this work is ignored, and thus not used to address the challenges of “field science,” when it would be quite fruitful to do so. Indeed, Boumans does not relate his discussions in chapters 5 and 6 on the role of clinical judgments and expert consensus to this literature—there is not a single reference to the works by Pierre Duhem or Richard Rudner, to name two classic authors, and there are also no references to more recent work by Philip Kitcher, Heather Douglas, Helen Longino, Alison Wylie, or others who have been very productively commenting on problems related to underdetermination, inductive risk, and the role of values in scientific inference. One way to read these latter contributions is precisely that what one might be inclined to associate with the world “outside” the laboratory is both needed and unavoidably permeating the “inside” of the laboratory anyway: subjective judgment and ethical, religious, political, or social values are all entangled in scientific inquiry. These are exactly the aspects of field science that Boumans emphasizes as well.

More specifically, in the contributions of Pierre Duhem, the problem of underdetermination has been conceived of as a problem of laboratory science, to use Boumans’s terminology. Underdetermination, in turn, has led many commentators to consider the role of values in science (such as in the early contributions by C. West Churchman and Richard Rudner). There is no reason to believe that both the problems and their attempted resolutions in these discussions do not apply to science “outside” the laboratory as well. If theories are underdetermined by the available evidence in an environment that allows the scientist to control many relevant factors, then there is no reason to believe that they are not underdetermined if there is less control. The same goes for values: it is likely that the room and need for value judgments of scientists actually increases with decreasing possibilities of controlling factors. Indeed, most cases that are studied in the more recent literature on inductive risk refer to scientific activities outside the laboratory (such as activities closely related to regulation and policy advice). We might of course uphold the distinction between science inside and outside the laboratory for other reasons (e.g., because we endorse a strong dichotomy between the natural and the social sciences). We might also conclude that there is nothing in the distinction that requires us to proceed without taking the literature on underdetermination, inductive risk, and values into account (so there may be a project in relating Boumans’s account to these debates). Yet, the fact remains that one of the most active debates in recent philosophy of science that deals with problems that are very close to Boumans’s concerns cannot be used to motivate his distinction and does not seem to be easily reconcilable with it.

The second worry is that the distinction between “inside” and “outside” the laboratory does not seem to relate well to the subject matter it was conceived of for—namely, measurement and expertise. Consider recent developments in metrology. It used to be the case that fundamental magnitudes were defined by artifacts: thus, a definite prototype of a meter bar or a kilogram were safely stowed away in ideal conditions and used to calibrate other objects in order to arrive at precise measurements. The artifact-based definition of fundamental units has been replaced by using nature’s constants (e.g., defining 1 second in relation to cycles of the radiation from a particular Caesium-133 transition). There has thus been a move away from artifacts to nature’s constants in the measurement of fundamental physical magnitudes, such as length, mass, and time. Now, it is not quite clear whether measurement via artifacts and measurement via nature’s constants do in any way map onto being field and laboratory science, respectively. However one may view this, surely the idea of defining fundamental physical magnitudes by nature’s constants is an instance of laboratory science: given the right conditions in a highly controlled environment will yield measurement, such as in the case of an atomic clock that aims to measure time. Yet, as detailed in an article by E. Tal (“Making Time: A Study in the Epistemology of Measurement,” British Journal for the Philosophy of Science 67 [2016]: 297–335), arriving at official time does involve the coordination of time measurement by atomic clocks on several continents. This, in turn, does involve decision making in expert committees, such as about uncertainty concerning the deviations from the idealized assumptions of time measurement in the different atomic clocks, and to reach consensus about how to aggregate the measurements into official time. Decisions about whether to introduce leap seconds into official time require similar judgments and discussions. It is not altogether clear how such aspects of laboratory science are fundamentally different from what happens “in the field” or “outside the laboratory.” On the contrary, it seems that there are similar problems in all the different ways in which physical and social magnitudes are being measured. More specifically, much of what is discussed in chapters 5 and 6 regarding consensus and judgment is inherent in any measurement. It is thus an open question to me whether Boumans’s distinction will advance systematic thinking about the challenges of measurement across the sciences.

When engaging with economic scholarship historically—which is what happens on the lion’s share of pages—this book is a wonderful resource. The methodology of field science promised in the beginning of the book I found to be, by and large, implicit in the treatment of the historical episodes and not developed explicitly enough. It is thus much harder to come to terms with and especially hard to relate it to scholarship in the philosophy of science. It should be noted, however, that the starting point of Boumans is rooted in a promising strategy that many scholars working on measurement will agree with: that different approaches within the theory and philosophy of measurement are best combined to capture all relevant aspects of measurement. The challenge to all scholars working on measurement lies in developing such accounts in a way that speaks to the problems faced by both natural and social scientists, whatever their prevalent methods and whether they operate primarily inside or outside the laboratory.