An alternative to Mendelism: W.F.R. Weldon's emphases on variation, ancestry and the environment

In the spring of 1900 three European botanists published papers reporting not merely the patterns Mendel had found in his hybrid pea lineages but also his explanation for those patterns. That striking coincidence brought Mendel's name – hitherto known only to specialists in hybrid plants – into general biological discussion, where two English biologists, William Bateson FRS and Walter Frank Raphael Weldon FRS, took notice.Footnote ² Bateson, based at Cambridge (where he had studied biology alongside Weldon in the 1870s and 1880s), rapidly persuaded himself that Mendel's work was fundamentally important and deserved pride of place in a new, quantitative, experimental science of inheritance. Under Bateson's energetic leadership, a growing corps of self-identified ‘Mendelians’ made it their mission to extend Mendelian analysis ever further, developing its theoretical principles and its practical applications, especially in agriculture but also in medicine. Bateson's coinage of 1905–1906, ‘genetics’, eventually displaced ‘Mendelism’ as the name for the new science.Footnote ³

While Bateson was transforming himself into Mendel's greatest disciple, Weldon, who held a chair at Oxford, was transforming himself into Mendel's (and Bateson's) greatest critic. Early in 1902, in a journal Weldon had helped to found, Biometrika, devoted to ‘the statistical study of biological problems’, he published an article now best remembered as the first to show that Mendel's data examined statistically are too good to be true. But for Weldon, as has become clearer thanks to recent studies of his unpublished writings, that empirical problem was but a symptom of a much larger conceptual one. Everyone will recall from school biology that Mendel dealt with binary characters – with, say, the colour of the pea seed as either green or yellow, the seed shape as either round or wrinkled, and so forth. According to Weldon, no other breeder's peas looked like that. In the article he included a coloured photographic plate showing pea seeds he had collected from commercial pea breeders in Britain and France. Yes, some peas were unambiguously green, and some unambiguously yellow. But between those extremes were varying shades of greeny yellow to yellowy green: a colour continuum.Footnote ⁴ The lessons he drew for inheritance and its study were threefold:

1. 1 Variation matters. It matters that real peas are not just yellow or green, and are not just round or wrinkled. Actual variability should not be idealized away, by the use of simple categories such as ‘yellow’ or ‘round’ treated as descriptions of reality. Statistics is the descriptive language appropriate to biology because biological populations are always variable and statistics captures that variability. (Hence ‘biometricians’, the name by which Weldon's side of the debate over Mendel's work became known.) And because that variability might itself turn out to hold indispensable clues to how inheritance really works, a science of inheritance that deliberately disguised variability, that encouraged the forgetting of it, was not moving in the right direction. Or so Weldon thought.Footnote ⁵

2. 2 Ancestry matters. The particular ancestry, the particular lineage, that these peas come from, as distinguished from those peas, can matter. Again, from Weldon's perspective, the trouble with Mendelism – Mendel's paper unjustifiably blown up to a biological world view – was that it encouraged ignorance and incuriosity, in this case about the deep ancestry behind any particular individual's characters. According to the Mendelian, as long as true-breeding green-seeded peas are producing green-seeded peas, that was all you needed to know about them. Weldon, impressed by Francis Galton's teachings about the long reach of ancestry, thought that was a mistake, and one connected to the Mendelian mistake about variability. Behind all of the variability that the Mendelians ignored might lie the different kinds of ancestry that the Mendelians ignored.Footnote ⁶

3. 3 Environment matters. As Weldon saw it, from within a lineage, an individual inherits not just this or that ‘factor’ or ‘determinant element’ or ‘gemmule’ or whatever but a whole suite of them, and – crucially – these constitute a context, an environment, conditioning the visible effects of any one of them. Change the environment, and you can change the effect. In a book-length ‘Theory of inheritance’ manuscript that Weldon worked on throughout 1904–1905, the conditioning role of environments – not just cellular but more generally physiological and physico-chemical – was a major theme. Indeed, Weldon took that theme to be one of the main conclusions of experimental embryology in its glory period towards the end of the nineteenth century, and so, surely, something that any science of inheritance worth having had to take into account. In trying to articulate that science himself he had avoided the language of ‘acquired characters’, as he explained in the manuscript, because no character is either all-acquired or all-inherited, but all are always the joint product of what was inherited and what was around it.Footnote ⁷ Furthermore, and again, he took this contextualist message – and the related ones about variability and ancestry – to be thoroughly Galtonian. Our culture remembers Galton, coiner of ‘eugenics’, as a doctrinaire hereditarian, adamant that nature beats nurture across the board. But that was not at all Weldon's reading of Galton. And Weldon was probably the most careful reader of Galton there ever was.Footnote ⁸

The upshot for Weldon was that the Mendelian picture of ‘dominance’ as absolute – so, for example, of yellowness as dominant to greenness, no matter the context – was deeply misleading. In Weldon's view, to run a Mendelian experiment was deliberately to exclude all of the variability that would otherwise produce different kinds of pattern. If an experimentalist were so minded, he thought, a race of peas could be established in which greenness was dominant to yellowness. It all depended on the choices made, the contexts built. And to declare one pattern the natural one, and other patterns as somehow deviant, was just arbitrary. No, what biologists needed was a concept of dominance which treated it as context-dependent.Footnote ⁹

Weldon arrived at this perspective in the course of an increasingly bad-tempered debate with Bateson and his allies. Now known as the ‘biometrician–Mendelian controversy’, it ended only when, in the spring of 1906, Weldon – the leading biologist on the biometrician side, though his friend and collaborator, the London-based mathematician Karl Pearson, has got more historical attention – died unexpectedly after a short illness.Footnote ¹⁰ Mendelism went on to become one of the great success stories of twentieth-century science. In 1910 in Brünn, Bateson watched the dedication of a statue of Mendel installed in a square next to his monastery.Footnote ¹¹ More than a hundred years later, in classrooms around the world where the subject is genetics, students early on are taught about Mendel and his experiments crossing peas. In a well-constructed introductory genetics course at university level, students are inducted by degrees into ever more complex variations on the basic Mendelian example, in all the ways that Thomas Kuhn has made familiar to generations of historians of science. Genetics has moved on in all sorts of ways since the early twentieth century, but instruction in genetics, once organisms are in the picture, keeps faith with its beginnings, making room for later innovations in a structure whose core remains forever Mendel's.Footnote ¹²

That is how it happened. But was it bound to happen that way? Was it a foregone conclusion that our biology would, somehow or other, at some time or other, recognize and enthrone something like the Mendelian gene? Or, on the contrary, might Weldon's alternative biology of inheritance, so insistent on the conditioning role of context, been the success story? Could Weldon's account have been at least more successful than it actually was, had circumstances somehow been different, more favourable – if, say, Weldon had been more adept than he was at recruiting talented and zealous students, or had simply lived to finish and publish his ‘Theory of inheritance’? Supposing history had taken a different, Weldonian turn in the early twentieth century, might we now have a Weldonian biology of inheritance, just as successful as the Mendelian one became, but with a different kind of intellectual complexion? And whether it might have or might not, why should we care?

Counterfactuals and the history of science

I come now to my title, ‘Experimenting with the scientific past’. What I have just wondered about is what has come to be known as ‘counterfactual history’: what did not happen but might have. We can describe such wonderings, if we want, as experiments – experiments of the imagination or, if ‘imagination’ talk seems somehow un-serious, conceptual experiments, thought experiments. In the laboratory of the mind, we change something in the actual past and reason through the consequences. ‘Altered pasts’ is how the eminent historian of modern Germany Richard J. Evans labels the results. If that phrase sounds a touch sinister, affiliating the counterfactual historian with totalitarian rewriters of history, that is in keeping with Evans's disapproving attitude towards the genre. His 2014 book Altered Pasts: Counterfactuals in History declares counterfactual history fit for examination mainly as a historical phenomenon in its own pathological right – a sign of intellectually and politically unhinged times, when people who cannot face reality console themselves with wish-fulfilling fantasy.Footnote ¹³

We need a less polemical guide in thinking about counterfactuals-as-experiments. I find Ian Hacking's Representing and Intervening endlessly stimulating on the subject of experiment, and Hacking's writings overall salutary in the care he takes to be attentive to when words are being used literally and when not. When we theorize, we represent; when we experiment, we intervene. It is not literally the case that we can experiment with the scientific past in the way that we can experiment with, say, water. Water is around, accessible, available for scientifically motivated interventions (removing dissolved impurities, running an electric current through, and so on). The scientific past is not like that, not even remotely. It is completely inaccessible, un-intervenable-upon. So, if we are being experimental when we go counterfactual about the past in the laboratory of the mind, those experiments are not literal experiments in the way that watery interventions are.Footnote ¹⁴

Fussiness at this stage will pay dividends later, when I will recount a literal experiment that does bear, I contend, on our knowledge of the counterfactual scientific past. But even that will fall short of the standard of a bona fide, literal-experiment intervention in the scientific past. As for not-even-remotely-literal experiments with the past, of the conjectural kind that exercise Richard Evans – what if Britain had not entered into the European war in 1914, and so on – we can describe them, if we like, as thought experiments in the laboratory of the mind, or ‘virtual history’ (Niall Ferguson), or ‘replays of the tape’ (Stephen Jay Gould). And there may well be some intellectual mileage to be gained in regarding the operations that these label as continuous rather than discontinuous with what happens in a literal experiment. But counterfactual conjectures are not in themselves experimental interventions.Footnote ¹⁵

What about counterfactuals and the history of science in particular? I think the most important points come down to three. First, counterfactuals are pervasive, like it or not, because historical explanation is a counterfactual business. Second, counterfactuals can express basic visions and divisions about science (and its connections to history, knowledge and reality). Third, the epistemology of counterfactuals – why we so often fictionalize in order to get closer to the facts – is mysterious. Let me expand on each of these in turn, before coming to what I take to be their overall thrust.

Experimenting with the Weldonian biology that might have been

Here we come to what, in methodological terms, seems to mark a difference, and for the better, between counterfactual history of science and counterfactual history of pretty much everything else. In articulating the contrast, we can do worse than return to Hacking's terms, representing and intervening. If I am curious about what might have happened had, say, the losing side in a battle or election been the winner, the methods at my disposal for satisfying that curiosity are all, in one form or another, representational. I can think about it, write about it, maybe even program a computer to run a simulation. I cannot actually go back and meddle. But if I am interested in what might have happened had the losing side in a scientific debate been the winner, or even just not as big a loser as it actually was, there is nothing in principle preventing me from undertaking an intervention: the bringing back into being, now, of that losing-side science.

Of course, a host of abstract difficulties and puzzles immediately present themselves. But let us put those aside for now and see how far we can get with a little concrete action.Footnote ²⁷ The losing-side science that I wish to concentrate on is one I introduced earlier: Weldonian biology. I take its intellectual essence to be well expressed in the following sentence, referred to earlier, from Weldon's ‘Theory of inheritance’ manuscript:

I cannot, anytime soon, usher into existence a Weldonian world, full of laboratories, fields and clinics that are themselves full of biologically and biomedically trained people who approach inheritance in such a thoroughly interactionist, nature-and-nurture-all-the-way-down manner. But the materials that might educate such people are another matter. A Weldonian curriculum in introductory genetics could be developed and then – passing from representation to intervention – taught to students. To get them pointed in a Weldonian direction from the start, the first meeting would dwell not on Mendel's pea experiments – the traditional ‘genes-for’ curriculum – but on an example where the effects of genes have to be understood alongside the interacting effects of many other causes – as with, say, the healthiness or otherwise of a human heart. From that beginning, the teacher, the readings and the associated activities would seek to press home, at every possible opportunity, the message that genes have the effects that they do on bodies only in developmental and, more broadly, environmental contexts. Where the traditional curriculum typically treats the Weldonian emphases on variation, ancestry and environments as largely ignorable luxury items, the Weldonian curriculum would treat them as primary. It would likewise dispense, as far as possible, with ‘dominance’. And when the time comes, eventually, to introduce Mendelian patterns, the students would learn to regard these not as showing inheritance at its most basic – as the great generalizations – but as special cases, exceptions to the rule, interesting precisely because, for explicable reasons, they can be understood as if genes acted independently of their contexts.

Suppose one could actually teach this curriculum, and get it, as pedagogy, working well enough that the students would not simply rush for the exits in dismay and confusion after the first meeting, but would instead stay the course, right through to the end. What would these students be like? In particular, how different would they be when it comes to holding that exaggerated view of the power of genes that our academic culture calls ‘genetic determinism’? The most exciting possibility, of course, is that these students would be interestingly different, and in a particular way. They would be less prone than students coming out of a traditional Mendelian course to accept unquestioningly the latest news reports about ‘genes for’ autism or schizophrenia and so on. They would be more sensitive to the ways in which individual differences can really matter in whether a BRCA1 mutation or other ‘disease gene’ mutation will lead to illness in any particular individual. They would be more willing to demand more information from the experts, not just about genes, but about the genetic backgrounds and range of environments in play in the research warranting claims to know what genes are ‘for’. They would, in short, be less determinist.

From early days, Mendelism and a kind of simple-minded determinism travelled hand in hand. I was pleased, in preparing this address, to find an example relevant to my Welsh setting, in Ted Porter's marvellous biography of Karl Pearson. During the Great War, Pearson kept a diary in which he recorded an encounter with a Mendelian. On Pearson's exasperated account, the latest hypothesis from this permanently misguided fellow was that ‘music in the Welshman was a “Mendelian recessive”’. That kind of reductive determinism incensed Pearson.Footnote ²⁹ After the war, the Mendelian-in-chief, Bateson, recalled the response of a Scottish soldier after listening to a lecture of Bateson's: ‘Sir, what ye're telling us is nothing but Scientific Calvinism’. Bateson thought that was spot-on, even considering ‘Scientific Calvinism’ to be an apt title for a contemplated, but never completed, collection of his popular writings and addresses on genetics.Footnote ³⁰

In our day, textbook Mendelism–determinism continues, entrenched. At the research frontier in biology and biomedicine, meanwhile, and in historical, philosophical and pedagogical studies of these fields too, interaction is where the action is. Consider the New York City-based Resilience Project, which encourages healthy people to have their whole genomes sequenced on the expectation that some will be found to carry genes which, according to the textbooks, should have been lethal long ago. Identifying such people is a first step towards figuring out what else is going in their genomes, bodies, environments to make those ‘genes of doom’ (in Steven Pinker's phrase) anything but. The hope is that medical science will be able to draw on that new knowledge of genes-in-context in order to develop a next generation of diagnostics and therapies.Footnote ³¹ In a similar vein, a recently published survey article on the genetics of neurodevelopmental disorders notes that ‘while a mutation associated with hemochromatosis or breast cancer might have high expression in one particular pedigree or clan, that same mutation may have very low expression in another pedigree, clan or group of unrelated people’. Elsewhere in this paper we find an explicitly Weldonian conclusion drawn: maybe the time has come to ‘just discuss the expression of each trait in the context of a phenotypic spectrum, which is of course what led Walter Frank Raphael Weldon to establish the field of biometry’. One of the co-authors of this paper – and the Weldon fan – is Gholson Lyon, an unusual combination of human geneticist and clinical psychiatrist. In other ways, too, he is an instructive figure, to whom I will return towards the end.Footnote ³²

Closer to home in disciplinary terms, the interactionist case has been put most influentially by Evelyn Fox Keller. The title of her book The Century of the Gene, published in 2000, was meant not to praise but to bury. One hundred years after the Mendelian rediscovery, and with the Human Genome Project entering its final phases, the time had come, in Keller's view, to relinquish the gene concept as no longer fit for scientific purpose. Up to a point, the twentieth century had managed well enough with the notion of genes as little atoms of heredity. They marched through the generations heedless of context, their structures and functions neatly individuated and stable over time, their autonomy in bringing about complex traits taken for granted. But by century's end, the gap between that notion and what was actually known had grown intolerably large. If the twenty-first century is to realize the potential of the genomic age, it needs to find new concepts and new language, better matched to biological reality (though quite what those concepts and that language were Keller did not say). Jenny Lewis, a specialist in genetics education, brought a similar message to those involved in teaching genetics in school. In 2011, Lewis warned that teachers who, not wanting to confuse students, rely on simple examples – where one gene produces one protein that in turn affects one character – risk pedagogical damage. For in so doing they may inculcate in their students ‘a deterministic view of genetics in which every characteristic is determined by a single gene’, thus making it harder for those students to absorb the interactionist message of genetics in the genomics age.Footnote ³³

By happy coincidence, Lewis turned out to be a colleague of mine at the University of Leeds; and together with a third Leeds colleague, Annie Jamieson, we got a chance to run a pilot-scale version of the Weldonian curriculum experiment outlined above. Our Genetics Pedagogies Project, as we came to call it, started in autumn 2012 and ended two years later.Footnote ³⁴ The basic design was to work with two groups of students, one taking Weldonian genetics (the experimental group), the other taking Mendelian genetics (the control group). Both groups were assessed beforehand and afterwards on their ‘belief in genetic determinism’, to use the phrase of the pioneer questionnaire-makers in this area. Under ideal conditions, the project would have recruited from a single homogeneous pool of students, assigned students to groups at random, used the same teacher for both groups, offered those groups courses having the same number of lectures of the same length delivered in the same style, and so on. Alas, our project did not take place under these conditions. Our experimental group consisted of twenty-eight humanities (mostly philosophy) students at the start of their second year, while our control group consisted of the same number of first-year biology students. Where the students in the experimental group met for nine extracurricular one-hour lectures taught by a member of the project team, and had associated reading and blog tasks to do, the students in the control group met for more than thirty credit-bearing lectures, in a large lecture hall with hundreds of other students, were taught by a member of the School of Biology, and had associated reading plus labs and problem sets. The incomparabilities introduced by all these differences are, to put it mildly, considerable. Even so, the ‘before’ data on genetic determinism from both groups showed that all the students entered their courses with more or less the same attitudes towards genes and their determinative power.

Annie Jamieson wrote the Weldonian course, taught it, and collected all of the relevant data. She knew Weldon's work intimately, through collaborating with me on a scholarly edition of Weldon's ‘Theory of inheritance’ which we expect soon to send to press. Better still, she is a developmental biologist turned historian of science, and a truly wonderful teacher. Her teaching materials, including her lectures in the form of audio recordings coupled to slide presentations, will soon be freely available online at the project website.Footnote ³⁵ For present purposes, a brief and imperfect overview of her first lecture, on the question ‘what is genetics?’, can serve to give at least a feeling for what it is like to be introduced in Annie's classroom to Weldonian genetics. She starts with inheritance as we observe it in the relations of parents and their offspring. Humans have human babies, horses have foals, tigers have baby tigers, and so on. There is always a sameness that gets preserved, and yet there is difference as well, and different degrees of difference. A baby tiger is much more like its parent than a foal seems to be like its parent. The children in a human family more closely resemble their parents than they will resemble the parents of other children, but are nevertheless different from their parents, and different from each other. To be interested in inheritance scientifically is to want to understand all of that: the sameness that gets preserved over time and the differences too.

From there we move to consider inheritance and development. In typical biological educations, genetics is handled in one course and development in another, separate course. In our Weldonian classroom, by contrast, students encountered genetics from the first as fundamentally tied up with development. Mature characters are not what are inherited. What are inherited are genes plus cytoplasm and, more generally, a developmental context. To think about how genes affect bodies is to think all the time about the developmental context in which genes function. That lesson gets affirmed and enlarged with a look at organism and environment, and an invitation to the students to think expansively about environments. These aren't, on reflection, just one thing, but many and diverse kinds of thing. Your family is an environment of a sort. So is the climate. Your diet. ‘What are your environments and how do they affect you?’ While the question lingers, two examples of genes and environment in interaction are introduced. In one the same genome with different environments produces very different results (queen and worker bees are genetically identical but the queen develops by being fed exclusively on royal jelly). In the other, different genomes in the same environment give different results (mice given access to the same level of food get fat or not depending on whether they have a mutation in the Ob gene, which, in its functional state, encodes a hormone called leptin, regulating appetite and energy expenditure).

I admire, and would like to think Weldon would admire, that pair of examples as a vehicle for impressing upon students a sense of genes and environment not just as interactive but as symmetrical. Genes here are not presented as super causes – what inheritance is ‘really about’ – with everything else involved playing a supplementary role. Genes are there in the mix, making things happen, but along with lots of other things. If we skip ahead to the very end of the lecture, we come – as in my anticipatory sketch – to the heart, specifically to cardiovascular disease and the question of the contribution that genes make. One by one, causal lines of influence are traced. Yes, genes can affect cardiovascular condition directly. But they can also affect blood pressure, the body's activity levels and a range of other factors, themselves often affected by non-genetic factors (such as smoking). The students then get an image representing all of those causal lines put together. It is as tangled a web of influence as could be. And where in there, the students are asked, is the ‘gene for’ cardiovascular disease? A hopelessly inapt question, as, we hope, they now see too.Footnote ³⁶

To our delight, almost all the students who began the experimental Weldonian course stuck with it, and we were able to gather ‘after’ data on their levels of genetic determinism, and to compare and contrast the results with ‘after’ data from our control group. What we found was that students coming out of the traditional Mendelian course were just as determinist about genes at the end as they were from the start, whereas students coming out of the Weldonian course showed reduced belief in genetic determinism, and to a statistically significant degree. Mindful of the small numbers involved, the previously noted imperfections in the experimental design, and Mendel's own difficulties with exaggerated statistics (and Weldon's role in first publicizing them), I don't wish to make too much of statistical significance here. But within the limitations of the study, these are encouraging results.

Making alternative pasts present: science as complementary HPS (and vice versa)

There are, I am conscious, too many dualisms for comfort in the above, most egregiously inevitability/contingency, but also representing/intervening, tradition/social construction, winners/losers, (f)actual/counterfactual, and science/history of science. All are ripe for reflection and revision.Footnote ³⁷ That is work for the future. For now, and in closing, I want to consider what might lie ahead for just the last pair on my list, science/history of science.

I begin with an anecdote. The Genetics Pedagogies Project was up and running when we became aware that Rosie Redfield, a microbial genetics specialist based at the University of British Columbia, had published a manifesto for a new genetics curriculum. She called for dispensing with the accretions of history – Mendel's peas included – in order to give students more directly useful and relevant instruction. Yet, for all that Redfield's project and ours had broadly similar reformist ambitions, her curriculum is not in the least Weldonian. Developmental context, and the Weldonian emphases on phenotypic variability, ancestry and environment generally, are in no way prioritized in her Useful Genetics course.Footnote ³⁸ It turns out, then, that a desire to update not-fit-for-purpose introductory teaching materials in genetics is probably a necessary but not a sufficient condition for revising those materials in a Weldonian direction. To acquire that distinctive ambition, with a view to diminishing support for genetic determinism, it helps – maybe it is even indispensable – to have spent time in the intellectual company of someone like Weldon, educated before Mendelism took hold. A new option for present-day science thus seems to depend on insights available now only from scholarly study of the scientific past.

It is worth stressing that our Weldonian curriculum does not represent anything like a total replacement of the existing curriculum. Rather, it seeks to take what is peripheral in the existing curriculum and make it central, and to take what is central and make it peripheral.Footnote ³⁹ My own inspiration for asking about reversals of this sort comes from a work of counterfactual history of science: a 2008 essay by Peter Bowler on the history of science and society without Darwin's Origin. The essay was a first statement in miniature of what became Bowler's Darwin Deleted. But what perhaps came through more clearly in the essay was the refusal of the notion that, when conjecturing counterfactually, there are only two possibilities: either everything would have been more or less the same, or everything would have been more or less completely different. In the Darwinian case, Bowler reckoned that, without the Origin, biology would ultimately have come to embrace roughly the same elements, but that these elements would have come about in a different order. This different order in turn would have had consequences for the relationships between and among the elements: for what came to be considered primary and what secondary, what mainstream and what marginal, what consensual and what controversial. Our Weldonian curriculum, and the experimental project elaborated around it, expresses that same willingness to contemplate a counterfactual past whose relevance, and maybe also knowability, lie precisely in not being that divergent from what actually came to pass.Footnote ⁴⁰

Taking seriously the value of such modest counterfactualism may hold the key to extending what we have done to other areas of science and history of science. As a case in point, consider Hasok Chang's Inventing Temperature and associated work, including experimental work, on boiling point. Back in the eighteenth century, it was taken for granted that water boils over a range of temperatures, depending on the condition of the water, the container and so on. Thermometers of the day thus represented boiling point as a spread around 208 degrees in Fahrenheit's scale. Nowadays, of course, boiling point is thought of as constant.Footnote ⁴¹ That shift, from sensitivity about context-dependent variability to indifference to it, has obvious affinities with the shift that concerns me in the history of genetics. Unsurprisingly, I suspect that the best way to characterize the chemical shift is not in terms of a forgetting about variability – as Chang has sometimes been inclined to characterize it, prompting objections from chemists – but rather a shunting of knowledge about it to the margins. Here it lies accessible in principle without in practice being assimilated into everyday chemical thinking and doing. Likewise, I reckon that a perspicuous way to develop Chang's conviction that this shunting had consequences for chemistry and chemists (and that both are the poorer for it) is to develop and teach an experimental chemical curriculum in which boiling point and other chemical phenomena are presented as variable. One could then see whether the students thus taught are interestingly different from students taught the traditional, constancy-emphasizing curriculum. The results may be as interesting for chemistry educators as for historian–philosophers of chemistry. More generally, this kind of endeavour presents an opportunity for a new kind of mutually beneficial collaboration between science educators and historians and philosophers of science.Footnote ⁴²

Inventing Temperature ended with a vision for history and philosophy of science (HPS) as ‘complementary science’. For the sake of the advance of scientific knowledge, historians and philosophers of science should work in close proximity to scientists, not actually in the lab but just down the corridor. Here, their investigations into neglected phenomena, debates shut down too soon, and so forth, might serve the cause of creative science by providing just the right measure of provocation.Footnote ⁴³ If I had to give a programmatic name to our experimental curriculum project, it would be Chang's in reverse: ‘science as complementary HPS’. The aim in the first instance was to find out more about the history of science and science education. How much did the biometrician–Mendelian debate really matter? Was a critical perspective like Weldon's bound to become marginal, whatever the historical circumstances that in fact made it so, because there is some intrinsic flaw in it, rendering it incapable of sustaining successful science? If so, then – recall Macaulay – the particular marginalizing circumstances are of relatively little interest, because, if not for them, there would have been others, and the end of the story would have been the same. But if, on the contrary, and as our experimental results suggest, there is no intrinsic flaw, then there is a far more compelling case for concentrating on the particulars of the biometrician/Mendelian moment in explaining why Weldon's perspective got marginalized as and when it did. Add in the apparently better match between twenty-first-century biology and a Weldonian curriculum over a Mendelian one, and the problem of explaining the Mendelian triumph – and the need for a better explanation of it – only increases. Supplying that explanation will be one of the goals of my book in progress.Footnote ⁴⁴

But I gladly fly the flag for HPS as complementary science, too, and have been pleased that, through the magic of social media, my counterfactual inquiry into the Weldonian biology that might have been now shares an electronic corridor with Gholson Lyon's lab at Cold Spring Harbor. Lyon found out about the Weldonian research through a blog posting from just before we launched the Genetics Pedagogies Project, and has been cheering it on ever since. In the article of Lyon's from which I quoted earlier, Weldon provides the first epigraph, taken from the 1902 paper with the variable-peas photograph:

The article closes with a broadside against ‘the current paradigm of genetic determinism’, depicted as stretching back to the time of William Bateson, and castigated for inculcating indifference to variability, ancestry and environment. ‘Categorical thinking misses complexity’, say Lyon and his co-author, Jason O'Rawe.Footnote ⁴⁵

Weldon is emerging for some critically minded scientists in our own day as an ally from across the seas of time. What about Mendel? Some might complain that it is a poor anniversary gift to jettison him from his place of honour in the genetics curriculum. Let me suggest that grumbling along these lines, though understandable, is misguided. If we want to honour Mendel, then let us read him, and read him seriously, which is to say, historically, without back-projecting the doctrinaire Mendelism that came later. Indeed, in the service of these better readings of Mendel – and by way of a genuinely suitable anniversary present – the British Society for the History of Science will later this year be publishing online, in freely accessible, fully annotated form, a new, scholarly English translation of Mendel's pea-hybrids paper, by Kersten Hall and Staffan Müller-Wille. Read Mendel, but let him be part of his time. Likewise, let our biology students be part of their time, by giving them a genetics curriculum that is fit for purpose in the twenty-first century. If we ask them to read Mendel, we should do so not to fill them with slack-jawed wonder at Mendel's achievement, but to help them appreciate how even the most creative and rigorous science – and Mendel's was first-rate on both counts – shows the stamp of the historical circumstances of its making. To learn that lesson well about past science is to bring a welcome level of self-awareness and critical self-reflection back to the present.

The Mendel anniversary of 2015–2016 is among the circumstances stamped all over this address. I end with a tribute to a less conspicuous circumstance: the seaside city of Swansea where the address had its debut. The very possibility of a Weldonian biology was arguably born in Swansea, as it was there, at the annual meeting of the British Association for the Advancement of Science in August 1880, that Weldon – still a student at Cambridge – first saw Galton in action.Footnote ⁴⁶ The meeting did not go down in the annals of the BAAS as one of their great ones, as it was a little under-attended. But Galton was everywhere. He described some new mathematical instruments he had invented. He reported the work of a committee which he had overseen, charged with gathering and analysing anthropometric data. And he gave an evening discourse during which he talked about his then-new technique of composite portraiture, whereby he took photographs of individuals and merged them in order to discover what kind of a type lay behind them.Footnote ⁴⁷ Would Weldon have been as eager to read Galton's Natural Inheritance when it came out in spring 1889 had Galton not made the impression he did at Swansea? Inheritance wasn't what Weldon was working on, nor was his research up to that point even the slightest bit mathematical, let alone statistical. But he read it, was excited by it, and sought not only to carry out research in its image but to correspond with Galton about it. The rest is history, actual and, now, counterfactual too.