My undergraduate seminar in the history of science focused on the history of quantification the last time I taught. I really enjoyed the course, and will likely offer that focus again. Here is the course description:
We live in a world consumed by numbers. From astronomy to physics to actuarial analysis to weather prediction, complex methods of calculation dominate our modes of knowledge production and consumption. No body of evidence is thought to be more persuasive—or deemed more objective—than quantitative data. And more data: according to Wired magazine, we have entered the “petabyte age,” the consequences of which “force us to view data mathematically first and establish a context for it later.” This is a brave new world with longstanding and established antecedents we would do well to better understand.
This course means to trace the story of how one method of representation—numbers, data, and quantification—virtually rendered theory and interpretation irrelevant. How is it that western understandings of truth came to be intimately and exclusively connected to statistics and mathematics? Seminar readings will begin by surveying the manner in which numbers became the foundation for social conceptions of truth. The second portion of the course—and student research projects—will concentrate on issues concerning the impact of this quantitative turn as it has influenced efforts to monitor the environment and understand environmental hazards. How did quantitative data transform the study of ecology? How or why did risk and environmental policymaking rely so heavily on numbers and not other, qualitative factors? How have we interpreted parts per million and parts per billion as information pertaining to climate change, body burdens, or risk? And how have mathematical models shaped the science of environmental prediction? Underpinning each of these questions is a deeper social inquiry into the connections inherent in science, politics, and authority that invites investigation into the historical relationship between knowledge, expertise, and power.
Better than most of my course offerings—even those explicitly focusing on environmental history or the history of sustainability—the readings and investigations zeroed in on my specific research interests. I was also fortunate to have an exceptionally strong group of students, many of whom subsequently went on to do graduate studies. That helped make this a rich teaching experience.
But I bring this up not just to reminisce. Just the other day, I came across a particularly interesting and provocative read online, which offered eight of the most “shocking” prognostications about the future. This from the World Future Society‘s site, which is often worth a look. Recall Wired‘s ushering in the petabyte age, and consider Vinod Khosla’s prediction that “big data will replace the need for 80% of all doctors.” It’s worth reading beyond the link’s short synopsis of his assertion to get your head around this, but it does seem to be championing the advent of big data as already here. Which is maybe not news, though its repercussions will continue to startle and amaze. There are lots of ways to think about this. It reminds me of this interest in the history of quantification, but it also has distinct connections with the history of the future, too.
Almost as a sidebar, the other good number news is that McMaster University has just introduced a new combined honours program between History and Mathematics. Exciting opportunities lie ahead.
I seem to have developed a longstanding interest in the history of milk. Inadvertently. Unintentionally. And not really. Let me explain.
During graduate school, I wrote a seminar paper on the swill milk scandal as reported in the media in 19th-century New York, published, among other places, here. Later, my interest in Barry Commoner’s social and scientific activism provided me with opportunity to investigate the Baby Tooth Survey of the 1950s and 1960s, which tested the amount of Strontium-90 in baby teeth as a result of nuclear fallout. As a means of alerting the public the hazards of radioactive fallout, Commoner and the St. Louis Committee for Nuclear Information engaged in a public study of baby teeth. Strontium-90 was a radioactive and calcium-like by-product of the tests that traced a similar path through the biological food chain. Strontium-90 followed calcium from soil to plants to animals, and into human bodies. And especially young bodies through milk consumption, which has long been known to be good for bones, teeth, and hair. Strontium-90 doesn’t offer the same benefits. Instead, it concentrates itself in bones, exponentially increasing one’s risk of cancer. Reading mothers’ letters accompanying the baby teeth is quite a moving experience (these are housed in the Western Historical Manuscript Collection at the University of Missouri-St. Louis). Imagine discovering that the milk you made your child drink might actually contain radioactive poison; parents were terrified and heavily involved. So, for one reason or another, milk keeps popping up in my work. (Interestingly—or, maybe, not—much of my focus on milk stems from its consumption, when most of my research tends toward production questions). To make matters even worse, I am also supervising a dissertation on cheese, which will be excellent.
This is sort of a by-product of my focus on new technologies developing environmental hazards, and there seems to me to be no greater risk to public health than threats to our food systems. Too: this is likely coloured by my being a father, and the heightened risk to small, developing, vulnerable bodies being especially susceptible to contaminants. Milk, especially, is of vital importance to children’s diets.
At any rate, the above serves as context for my pausing to read through a recent Wall Street Journal article on the decline of milk consumption. Note that I run off on a tangent below and that the WSJ article serves only as a jumping-off point. It’s well worth the read, however, and posits some interesting explanations for the decline of per capita milk consumption. In addition, the decline of milk consumption introduces some intriguing retail dynamics. According to the WSJ, milk products were typically kept at the back of stores and were used as a loss leader, enticing customers to walk through the aisles. But, to me, the premise for the whole article was what caught me:
Per-capita U.S. milk consumption, which peaked around World War II, has fallen almost 30% since 1975.
The historian in me wants to trace this decline and explain it. I must confess to initially putting the cart before the horse; given my interest in the Strontium-90 scare of the late 1950s and early 1960s, I wanted to see if this had any impact on American milk consumption (this doesn’t interest itself in the more recent decline, which the WSJ tries to justify). That is to say, is it possible to link reduced milk consumption to the period surrounding the Baby Tooth Survey? The more dedicated historian would be inclined to investigate this data without any predetermined assumptions about what it might yield, lest s/he find ways to skew the data to generate the desired result. There are, after all, lies, damned lies, and statistics. But this is merely a playful exercise in playing with data and thinking historically about it, so I’ll forge on.
As dates for data analysis, let’s start with 1958 (when the Committee for Nuclear Information initiated the study) and conclude with 1963 and the signing of the Nuclear Test Ban Treaty. It might also be worth watching 1961, when Louise Zibold Reiss published the Baby Tooth Survey’s preliminary data in Science. Here’s the data:
Caveats: I chose—rightfully or wrongfully—to concentrate just on plain milk. Keep in mind that dairy—butter, cheese, yogurt, flavoured milk, etc—is a much bigger and more complex issue. And complexity is a key feature of doing good history. Also, we might examine the politics of breastfeeding over the past century to fully round out this study. Maybe these are important, maybe not. The historian and the focus of the project would need to determine scope and relevance. But cutting corners by skipping variables can be problematic.
Visual data provides exciting opportunity to discuss history. And it raises a number of questions. My initial “guess” as to why milk consumption grew so markedly in the 1940s involved the 1946 National School Lunch Act, which explicitly included milk (up to 2 pints per day) as a free school staple. But milk consumption peaked the year before The Act came into effect. In fact, it seems as though the Act precipitated the decline between 1945 and 1950. In the absence of other information, this provides students with the opportunity to infer what happened. Less milk purchased for home consumption? Or were other factors at play? Proper analysis requires developing a better and more nuanced timeline—reading behind the data to understand the social and cultural politics of nutritional science and how and when various dairy associations and councils lobbied more or less effectively. Raw data doesn’t explain this (though it might help tell us where/when to look for clues).
Another interesting feature of the general decline of milk consumption during the 1950s and 1960s is that it coincided with the square containers, which allowed for the more efficient sale of more milk (cartons made it easier to sell greater quantities of milk for less money than bottles). But still people were drinking ever less milk.
Back to my initial curiosity concerning the influence of the Baby Tooth Survey. It would appear as though nuclear fallout played a limited role. Here is the data on a year by year basis from 1945 to 1965:
The Baby Tooth Survey began in 1958. In 1957, per capita milk consumption was 290.4 lbs, down from its all-time high of 347.2 lbs in 1945. Which is to say that a decline had already occurred. Playing causation and correlation, it is interesting to note that after a relative levelling-off of milk consumption between 1948 and 1957, per capita consumption proceeded to drop steadily from roughly 1958 to 1963, the year the Nuclear Test Ban Treaty was signed. This is more evident in the smaller graph. Witness:
From a visual perspective, how you measure makes a difference. While the longer time span does not seem to suggest any correlation between the Baby Tooth Survey and milk consumption, a closer, annual look at the data indicates there could be some kind of connection. But the historian needs more data, more information, and more context. Where would you look for more clues?
It’s also easy to lose sight of what it is we’re examining. Note that this decline occurs during the Baby Boom. One feature of the Wall Street Journal‘s account of the contemporary decline in milk consumption involved an aging population. During the post-World War II baby boom, you would think that there would be an increase in milk consumption. But note: we’re looking at data for per capita milk consumption, which is to say: milk consumed per person. And this is where the Wall Street Journal trips up a little. Here’s a summary of milk production and sales over the same period as the first graph:
So milk production and sales have been steadily increasing (and, going back to the long-term consumption graph at the top, only post-Baby Boom, by the 1970s, did milk consumption drop below the fairly steady per capita consumption levels of the 1920s and 1930s). We might infer that increased industrialization and wartime expansion might account for the growth in milk sales and consumption (compare with the consumption table above—both note marked growth during the 1940s) during World War II. Of course, US population has grown steadily since World War II, which accounts for the decline in per capita consumption. But this would indicate that the Baby Boom likely did have an overall impact on milk sales. Curiously, though, the 1946 National School Lunch Act has no impact again, though you might expect to see some increase on this graphic. But we could see the influence of milk cartons in the growth in milk sales during the 1950s and 1960s. Perhaps. This is still rather speculative.
Indeed, none of this is designed to be conclusive; this is all more a think piece and brief discussion of how historians might use data, noting some of the prospects and pitfalls that students should keep in mind. In grading my last batch of history exams, I was struck by the number of students who clearly retained lots of course information, but seemed unable to process it and determine what information was important and what was not. As a result, they threw everything at the wall (or question) and hoped that some of it would stick. Without trying to sound condescending, this is unsatisfactory history. For me, the real pleasure in this kind of analysis is the joy of the hunt: trying to prioritize factors that help explain the numbers. History doesn’t stop at trends of increase or decline; instead it begins there. Historians need explain why events took place.
Historians cannot interpret results that the existing data does not support. But we also need to be careful to ensure that what we are looking at is an honest reading of what data tells us. Similarly, we must consider what we are looking for. I’ll follow up on this in the new year with some data on bicycles and bicycling as part of my bike history courses starting in January. It’s easy to get lost in graphs and data—I love looking at all kinds of statistics. But, increasingly, with more and more information available to the historian in the digital age, the onus is on us to ensure that we can not just present the data, but also explain it. That involves looking into and beyond the data sets.
On November 15, the McMaster Rolling Seminar: La Vie Vélo hosted a roundtable panel on coffee & bikes. Recently the Canadian Cycling Magazine posted a short video, consisting of interviews with a couple of the participants. You can link to their page here.
As we suggested at the time, there is a longstanding relationship between coffee and cycling, which has also been mentioned recently on the Inner Ring site, here. Two more pieces on coffee and cycling can be seen here and here. But also more. Here, here, and here.
(Photo taken at Café Domestique in Dundas. Note, not just the fine espresso and steamed milk my daughter enjoyed, but also the table, made out of a time trial wheel).
This morning, my students wrote the final exam for their History 2EE3 course. It’s a three-hour exam consisting of a short-answer portion (writing five terms from eight choices) and an essay question (four were distributed on the course study guide, and one showed up on the exam). Study guides can be daunting things. I provided close to 90 terms I expected students to know. The alternative was to not provide a study guide and expect students to be familiar with all the material covered, but I felt there was some pedagogical merit to them having a list of terms that might help them to organize their thoughts around course themes. This wasn’t designed as a shortcut study guide, but rather a method of helping students to pick up on the big-picture questions associated with the course. My weekend will be spent determining how well I communicated the course themes and ideas and how well students made the connections across time and place.
In a recent lecture on the history of the computer, though, I couldn’t resist giving the students inside help. On the slideshow’s opening page, looking a bit like part of the slide’s design, was a slew of binary (edited to add the screen capture and clarify the binary message):
If you’re familiar with binary, you might be able to read the the first box, which says “SHH! It’s a secret!” It was all a bit of play; the second box invited students who deciphered the binary to tweet “#2EE3 got it.”
Three students replied out of a hundred. One had his brother tweet for him inside of 48 hours. The second approached me after class a week later. I don’t know if these are good odds or bad odds, but it was a nice little inside game. A big part of the course had been the human affinity for order and patterns, and these perceptive students had picked up on the fact that the random sequence of 0s and 1s were not, in fact, random. What neither student mention to me, though, was that they had noticed similar messages coded into subsequent pages of the slideshow, which consisted of the terms that would show up on the final exam. For example:
Maybe they saw this and were able to render their exam preparation more efficient; maybe not. The third student tweeted on the eve of the final, claiming she had just become a lot less stressed about the morning’s final, which, I assume, means she cracked the code throughout the slideshow.
It was a good semester, though, and I enjoyed teaching my survey of science and technology in world history. It’s a fun course, and rewarding to see a swath of students from across the various faculties on campus. I’m always fascinated by how they take to the course in drastically different ways. I look forward to teaching History 2EE3 again next September.