It is certainly widely appreciated that there is much to be gained in the fertile crosstalk between science and history —— whether bringing a historical perspective into the science classroom (Wieder, 2006) or a scientific perspective to the study of history (McElvaine, 2002; Smail, 2008). Perhaps the major impetus for using history in teaching science has been to better transmit a sense of how science actually gets done (Conant, 1948; Johnson, 1996). I describe a different approach: using historical materials not from a history-of-science perspective but in actively developing students' analytical abilities with respect to scientific experiment and current scientific reports.
Over the past 10 years I have been giving a course for bio-education and upper-level biology students that uses selected historical materials as a starting point for developing students' critical capacities in approaching the modern scientific literature. The historical materials that form the first half of the course are selected not so much for their historical significance as for their ease of adaptation for student analysis. This generally means that they are concise and have clearly delineated experimental questions which are then expressed in a cogent logic of experimental design. In the second half of the course, the students' developing analytical skills are brought to bear on readings from the current scientific literature, often drawn from such sources as Nature, Science, Public Library of Science (PLoS) journals, or Proceedings of the National Academy of Sciences.
Why begin this active approach to scientific literature with historical publications? The expectation was that students might have a much easier time analyzing texts from an earlier period before proceeding to readings in the current scientific literature. For instance, one of the most difficult tasks in reading scientific literature is elucidating the assumptions being made (Walker, 2003; Yip, 2003; and some nice examples in Seethaler, 2009). Using reports from much earlier periods, in which the assumptions can be dramatically different from those of our own time, makes the very presence of assumptions more apparent (as illustrated below by a specific example). Also, historical papers are often based on standards of scientific proof that are very different from those we use today. The lens of time greatly magnifies these deviations from contemporary norms. But once students understand how to identify and analyze the reasoning that underlies an experiment, the process is readily transferable to current literature. This is also true of experimental design: many of the earliest historical examples come nowhere close to the stringency of modern standards of scientific controls. Seeing where controls are lacking, why that weakens the report, and what controls might have been used aids students not only in reading scientific literature intelligently but in designing their own experiments. Again, such skills are easily transferred to the contemporary reports considered later in the course.
The Saggi Di Naturali Esperienze, a Book from the 17th Century Accademia Del Cimento
The earliest report I have located for the purposes described above, and the one that I start with, is from one of the first formal scientific societies, the Accademia del Cimento, which was active in Florence from 1657 to 1667, predating the Royal Society of England (Ornstein, 1928; Waller, 1964; Middleton, 1971; Boschiero, 2007). The Accademia was founded under the patronage of Grand Duke Ferdinand II de Medici, who succeeded Cosimo II in 1620 as Grand Duke of Tuscany, and Ferdinand's brother, Prince Leopold de Medici. Such patronage gave the experimenters a degree of protection. This was, after all, but a few brief years after the trial of Galileo, and it was specifically to further the experimental approach to nature developed by Galileo that the Accademia was formed. Indeed, the Accademia was sometimes referred to as ““Accademia Galileiana del Cimento.”” The Medici had long been supporters of science as well as protectors of Galileo, so the establishment of this early scientific academy was well within the Medici tradition. The patronage also brought with it other considerable advantages: funding, the services of some of Italy's finest glassblowers for executing the experimental apparatus designed by members of the Accademia (see Figure 1), and an enviable venue for the Accademia's meetings —— Prince Leopold's Pitti Palace itself.
When Prince Leopold was made a Bishop in 1667, the society ceased its meetings and prepared a magnificent book of some of its experiments. The title, Saggi di Naturali Esperienze, can be roughly translated as Selected Natural Experiments, representing a tiny fraction of the experimenters' decade of work. The book is a magnificent folio measuring 14 ×× 9.5 inches, with numerous full-page renderings of apparatus, numerous woodcut designs based on the society's interests (Figure 2), and a decorated frontispiece (see Figure 3). The society's motto, ““probando e reprobando”” (test and test again), which can be seen in Figure 3, is actually a quotation from Dante, adapted to this new scientific context. One thousand copies of the 1667 first edition of the book were printed, but none were sold. All the volumes were presented by the Prince as gifts to foreign rulers and learned societies. A second edition of 1,000 volumes, virtually identical to the first, was printed in 1697, and again all were given as presentation copies. The illustrations shown in this article are taken from the second edition of the Saggi.
The experiments described below are from Richard Waller's 1684 English translation of the Saggi (Waller, 1964: p. 155). The original text as it appeared in the Saggi can be seen in Figure 4. It is a very small section toward the end of the book, translated by Waller as ““some EXPERIMENTS to know If GLASS and CRYSTAL be Penetrable by ODOURS and HUMIDITY.”” Thus, the experimental question is very clearly stated. I find it useful to first have students think about how they might design approaches to answering this question. It soon becomes obvious what components would be needed in any experimental design: certainly some well-defined source of an odor and, very importantly, an assay —— that is, some means for detecting an odor if it is actually present. In addition, there are considerations of the equipment that will be needed, such as glass or crystal containers, means for filling them, some idea of what volumes to use, the shape of the containers, and so on. Here, in Waller's translation, are the Accademia's experiments:
Oyl of Wax, Quintessence of Sulphur, and Extract of Horses Urine, which are reckon'd the most acute and strong smells that are, do not sensibly transpire through a Sealed Glass Vial, as could by many persons that tried it, be perceived, tho it was heated.
The Halitus also of that thin Spirit that flies away upon cutting an Orange, or Lemon Peel, or which in a small Thread spins out of the same Peel when it is squeezed, did not penetrate to give any smell to a little Water contained in a Cristal Glass Sealed Hermetically.
In like manner, Sealing up a Partridge in a small Glass Vessel, and setting it in a corner of the Room, and bringing a Setting Dog in, we led him round, near the place where it was set; but he shewed no sign of perceiving the Partridge.
The choice of odors selected for testing follows pretty much the same logic any investigator would employ: ““the most acute and strong smells that are.”” That is, acuteness and strength in relation to the assay employed —— in this case, people. Now, how good is that assay? At first it seems crude and unscientific. But why?
The answer, as in any evaluation of this sort, really comes down to missing controls. After all, what do we mean by ““odor”” —— other than a sensation that we are able to perceive in response to certain substances as they impinge upon the olfactory apparatus. But for the assay to be convincing, both positive and negative controls would have to be included. The former would assure us that, yes, if the odor were present, our assay would detect it; and the latter would assure us that if it were not present, it would not be detected. Students often struggle with how to develop the best such controls within the context of the overall experimental design. The key, as with any control, is to keep all factors exactly the same, with the exception of the component being measured. Thus, a positive control might be a glass vial that appears to be absolutely identical to the one in the experiment and also contains the exact same substances —— sulfur or horse urine. But in this case, the vial could be permeated with multiple small pores so that odor could readily escape. The question would then be whether or not the persons involved registered a positive response. This could be carried further by using less and less of the test substance to determine the assay system's limits of detectability. The negative control would use a similarly porous vial containing only water.
We can now proceed to experiment 2: the lemon peel. What was the purpose of this experiment within the logical structure of these studies? The experimental question was whether or not odors could penetrate glass, and the second experiment advances the scientific argument beyond the first by examining the opposite direction of penetrance. Experiment 1 says that odor cannot pass from inside to outside a sealed glass container, and experiment 2 says that it does not work the other way either: the water inside the vessel does not become impregnated with an odor that is present outside the vessel.
Finally, there is experiment 3: the dog and the partridge. How does this further the pursuit? By changing the assay. The detector is now a dog, perhaps an advance on the possibly subjective —— and uncontrolled —— human participants. But the odor's source this time is not so well described. Is this a live partridge? A newly killed one? A rotting carcass? And what, exactly, would set the setting dog off? There are obvious gaps in the design of this part, although it is a creative departure with a clear intent.
Students generally have a terrific time getting into this simple presentation and locating the gaping questions to be asked and pursued. But these, of course, are just the sorts of questions to be brought to the reading of any scientific report —— made more obvious thanks to the intervening 350 years. It is important to emphasize that these historical experiments were done by important and oftentimes brilliant investigators. Why they asked the questions that they did and how they pursued their investigations are not trivial matters. It is just science done at a very different time.
Looking at Assumptions
Every series of investigations has within it numerous assumptions. Not all of them are necessarily correct. However, since we generally share most of the assumptions made in contemporary literature, they often become extremely difficult to identify. Assumptions are especially important in what has become the prevailing modern sense that scientific experiments can be used to falsify but not to verify a theory (Popper, 1959; Magee, 1973). But, it has been posited, rather than modifying theory in response to apparently contradictory data, it is often possible to reexamine the assumptions of the original experiments and modify those assumptions to accommodate the new findings while retaining the original hypothesis (Harding, 1975; Norton, 2008).
So what assumptions are being made in the Accademia's report on odors? As in the Sherlock Holmes story ““Silver Blaze,”” assumptions generally reveal themselves as the dogs that do not bark. Thus, the Accademia experiments seem to assume that the nature of the venue —— indoors, outdoors, temperature, humidity, room dimensions, and so on —— does not influence the experimental conclusions. Also, the nature of the containers is ignored —— the type of glass used, its chemical composition, whether opaque or clear, its thickness, its shape —— all remain unspecified. There are also more fundamental assumptions being made —— for example, that odors really exist. There are also experimental assumptions that could be pursued —— is it possible that horse urine, for instance, modifies the property of glass, rendering it impermeable to odor? What about the time of year when the experiment was done, the time of day, and were the people doing the odor detecting male or female? What ages? What was their religion, their politics? Were they overweight, bald, tone deaf? Did they themselves have strong odors?
Perhaps we can dismiss most of the almost limitless list of assumptions as unnecessary for further consideration. But how can we be sure? Who, for instance, just a few years ago, realized that when reporting the size of an object they also needed to tell us how fast it was going? What about the horse urine —— do we need to know whether it was from a mare, stallion, or gelding, what the horse's diet had been, its age? Also, what if strong odors can't pass through glass but weaker ones can? Some subset of assumptions can be crucial for both limiting the interpretation of experimental results and understanding what further experiments might be meaningfully undertaken.
Continuation of the Approach & a Cooperative Proposal
Using the Saggi report as an easy entr, the course then progresses to the more detailed classic experiments of Pascal on the nature of the vacuum (Toricelli's 1644 letter to Ricci and Pascal's ““The Great Experiment on the Weight of the Mass of the Air”” —— sometimes cited as the first experiment with an intentionally designed control), selections from Harvey (both observational and, especially, his quantitative arguments —— mainly chapters 8——13 from De Motu Cordis), early studies on fermentation, and on through Bayliss and Starling's classic studies on pancreatic secretion (Journal of Physiology, 28, 325——353, 1902), Fleming's paper on penicillin (British Journal of Experimental Pathology, 10, 226——236, 1929), and, if time permits, D'Herelle's virus work and ensuing studies on bacteriophages and the origins of molecular biology (British Medical Journal, 2, 289——297, 1922, and later work from Delbruck). These are supplemented with readings from Francis Bacon's ““Idols”” (from his Magna Insauratio aphorisms and readily available online), excerpts from Claude Bernard's Introduction to Experimental Medicine (first published in 1865 and still readily available in the Dover paperback edition from which I use pages 6——26, 40——53, 151——163, and 178——183), and a selection of readings and demonstrations concerning cognitive illusions and errors (I update these regularly as new materials become available). Unfortunately, space does not permit a description of how I integrate these into the class —— however, the area is broadly reviewed by Pohl (2004). The Accademia kickoff reading seems to keep students nicely anchored as we proceed through these and on to selections from the current scientific literature. I have found class responses to this approach truly rewarding. Students seem to enjoy the process, appreciate the benefit of the tools they are exercising, and, in general, just have a really good time.
Finding appropriate reports, either historical or modern, for the sorts of exercises described is a laborious task. They have to be concise or capable of being made so through simple excerpting. The papers should also have clearly stated experimental questions, require no significant additional technical instruction, and exhibit a logical process whereby the experimental design corresponds to the initial scientific question. I invite teachers who are interested in pursuing this active use of scientific reports to suggest interesting materials in the biological sciences to add to my list. I would be happy to collect them via e-mail and then share them at large. I would also love to hear of experiences that other instructors have in using such materials, with the hope that they and their students will also find the approach dynamic, instructive, fulfilling, and fun.