Research Blog

Updated as the mood strikes me

  • ginamagostini

Casting bones pre-3D: A how-to-guide for misery

Updated: Oct 8, 2018

AKA: Behind the scenes of my dissertation

I’m a big thinker, by which I mean I like to design massive projects that speak to big, transdisciplinary concepts. This is why I’d never cut it in a field like molecular biology where one can spend an entire career studying a single gene and its associated regulatory elements. That work is super important, it’s just not for the perpetually-curious-yet-quickly-bored folks like me. Unfortunately, my “big vision” aspirations are not met with a commensurate level of “thinking it through” when it comes to logistics, the combination of which often means I design large, hopelessly complex projects requiring years of human labor. My girlfriend calls it ambition, I call it well-intentioned stupidity, but hey, as long as I stay productive (and sane), I suppose it’ll all be fine [1].

This is perhaps exemplified nowhere better than my dissertation and countless related side projects, the responsibility of which will likely be shouldered by my future grads. A behemoth, this project(s) ultimately required over $30,000 (a lot for lowly grads), 1,500 specimens, 10,561 little bone casts, and 62 x 1500 metric measurements, as well as nearly 10 years of my life that will never be returned to me (plus whatever reduction to my lifespan I’ll face due to the associated stress). It also proved that my superhuman ability to execute the same mindless task over and over [2] would’ve made Henry Ford propose were the temporal and, let’s say cultural [3], distance between us simply not so great. But before I get too far, I should probably start at the beginning—before I ended up spending every weekend night like this:

Like most things, it all started with an idea that sounded great on paper. It was 2008 and I was wrapping up my first major project, a forensics-y thesis investigating body mass and limb bone structure. Because I was young, idealistic, had no romantic prospects and also had no life, I used to spend my Saturdays lying on the floor of a tiny office, surrounded by a horse-shoe of printed articles [4]. By this point I’d already decided that I would pursue a PhD [5] and that I wanted to study something related to arm and leg bones. What that something might be, however, was undecided.

I read three papers that day which would solidify my fate [6]. The first article, by Ruff, Holt, and Trinkaus, focused on how arm and leg bones were modified by physical activity, something I already knew thanks to my thesis. The second article (technically a re-read, but that doesn’t sound as sexy) was written by Relethford and focused on how genetic relationships across human populations can be detected by looking at skull variation. The last by Weaver and von Cramon-Taubadel focused on a similar topic, but was an overview that presented genetic relationships in broader terms evolutionary terms, in particular what the skeleton can tell us about patterns of neutral evolution, which is a fancy way of saying variation due to things like descent, ancestry, migrations, and even natural disasters that wipe out parts of populations and any future genetic contribution those individuals might have made. It was a quote in this paper that gave me my “Aha!” moment [7]. I’ve translated it from academicese to the common tongue as thus (with some creative license [8]):

“Most studies of bone genetics focus on human skulls or teeth, probably due to the misguided assumption that only head bits are useful to study genetic signatures. However, nonskull bones are important too. Maybe they also have important things to tell us about genetics and evolution--we don’t know because no one ever studies them that way. Every once in a while someone does a nonskull study on how environmental factors (e.g., extreme cold) can change populations by weeding out certain individuals over time (AKA: adaptation via natural selection, you’re welcome high schoolers [9]). However, the extent to which nonskull bones reflect very basic genetic information hasn’t really been investigated in a formal way, and really probably should be before we write the whole thing off.”

“Huh. Well I should do that,” I thought. And so for my dissertation I decided I would target the diaphyses or shafts of arm and leg bones. I picked diaphyses not just because they were comfortable (I’d studied them for my thesis), but because they’re believed to be one of the most plastic parts of the skeleton. They do, in fact, change shape in response to physical activity [10]. But instead of studying activity like most other folks, I decided I would test whether these “very plastic” parts of bones reflected simple genetic relationships across populations, and, further, whether computer algorithms designed to extract genetic information from skeletons would be fooled when comparing closely related groups that had very different behaviors (like office clerks compared to factory workers). But this isn’t a blog post about my dissertation results (read here if you’re curious), it’s a blog post about bone casting, so let’s get back to that.

There are many ways to get shape data from diaphyses, and like most things in life it boils down to privilege. If you come from a program with lots of money or connections, you can use fancy things like laser or light scanners or even CT machines to scan whole bones. Then you have a 3D model of the bone and are pretty much set for life because you have nearly infinite datapoints at your disposal to be used on future projects. But if you’re a poor schlep like me, your options are more limited [11]. You can take external measurements of the bone’s diameter with calipers, but because external measurements are linear (two-dimensional), you can only approximate shape. If you want true shape, then you get to make bone casts.... many little bone casts that span the circumference of the diaphysis like so:

After crunching preliminary numbers, I decided I’d need data for at least 1200 people, maybe more. This would generate somewhere around 10,000 bone casts. “No problem,” I thought to myself. “I’ve got this.” [12]

Here are some things you should know about bone cast logistics before you ever make a single one:

  1. You know that stuff your dentist uses to make tooth molds? It’s called polyvinyl siloxane dental putty, and it’s what you use to make bone casts. It smells delightfully minty, which is no small gift when considering your primary workplace is a series of museum attics and basements.

  2. One large container of dental putty is sufficient to cast ~35 people’s worth of bones depending on how big they are. For the record, Saxons were biiiiiiiiiiiiiiiiiiiiiiiiiig. I’d collect data on ~300-400 people each summer, so I’d usually have to take about 10 containers with me.

  3. Dental putty is heavy. One entire 50lb baggage allowance went solely toward friggin dental putty. This means that the rest of the equipment (measurement devices, computer, etc.) *and* things like clothes had to fit into another single bag. Toiletries, make-up and other items to primp with? Forget it. You just have to buy all that crap when you get there or go au naturel [13]. But my mad packing abilities are a point of pride, especially given I’d routinely travel 2-3 different countries for 3-4 month stints each field season. Point in suitcase [14]:

BAM! And because this was when I went from England to 6 weeks in France and Italy, that bag also had toiletries, dental putty, licorice candy (which you can see took priority over the make-up bag), and a hair dryer in the same bag whaaaaaaaaat. Plus it met the draconian weight limits of the Euro Airlines to boot. So if ever there’s a reality competition show about packing, I’m going for gold. [15]

The actual process of making casts is both tedious and straight-forward. You orient the bone along the major anatomical planes, fix them in place with clay, lightly mark certain percentages of bone length so you’re casting consistent locations from one individual to the next, prepare the putty, make the cast, let it dry (but not for too long or you can damage the bone), label the cast, remove it and, voila, you’re done. I would repeat this process anywhere from 8-12 hours/day without anything beyond bathroom breaks because I’m a sadist who practices intermittent fasting so lunch isn’t really that important. For those of you visual learners out there, here’s a handy photo flowchart:

You rinse and repeat until you’ve either exhausted yourself or the patience of your curator, at which point you can go back to your hotel room, eat a wholesome meal of canned tuna, celery, and avocado and other nonrefridgoraterables, and probably drink (but not too much, because you have to get up early and do this all over again the next day)! By the end of the summer you have a few thousand bone casts, and at the end of several summers you have over 10,000!

This was the total yield by the end of the project—each box contains casts for 1-3 different samples. Only about half of this material was included in my dissertation. The rest is lying in wait until I have time to work with it.

Having all those casts is awesome, but guess what? They don’t tell you a darn thing except that you wasted a crapton of time making casts! That’s because you have to digitize them first, a process that requires trimming, tracing, and scanning every cast, and therefore takes as long as it did to actually make them to begin with (many thanks to undergraduate research assistants to helped to process some of the European material). Here’s a flowchart for that:

And once you’re finished with that, you have a bunch of digital renderings of your bone casts! Congratulations, you’re finally ready to do the sort of data analysis that would have been achievable on a machine running Windows 95!

And while there’s an understated beauty in a screen filled with high contrast bone cast renderings, they still don’t tell you anything. For that write a macro that pulls the images into an image editing software program equipped with a plugin that will take the image and extract information about the physical properties of the cast (shape, diameters in varying planes, circumference, areas) as well as key biomechanical information about the bone it came from (it’s inferred rigidity and strength under varying loading scenarios, like bending, twisting, compression and stretching). This is obviously a time consuming process, so I’d often just load a folder with a few hundred prepped pictures, launch the macro, and let it run overnight, praying to the computer gods it wouldn’t error out. I quickly learned it was best to bypass my computer entirely each morning until I’d consumed a sufficient amount of coffee to offset the inevitable frustration brought on by error windows. And in a testament to their incredible empathy, both of my cats, dog, and then S.O. quickly learned it was best to bypass me in the mornings altogether. The chinchilla, being perpetually ill-willed, was terribly unsympathetic to my struggle.

But with any luck I’d wake up each morning, meditate in front of the Bialetti, down 4 espresso shots, and wander upstairs to find a long text file of data that could then be shunted over to Excel. Every image successfully processed would return 32 beautiful, albeit completely indecipherable, numbers that would only make sense thanks to statistical models some time later. With 14 images per person, that’s a lot of numbers [16].

Once they’re digitized, you’ll never need the casts or their tracings again (unless you suspect you’ve found a labeling error and need to go back to the source). But hey, at least you have this nifty binder to carry around with you for life. One should never under-appreciate the utility of a good doorstop.

So after several more months, I finally had data [17]! But they were raw data, meaning they couldn't (well, shouldn't) be used yet [18]. First they must be size- and mass-standardized to ensure results generated later aren’t simply a product of the fact bigger people have beefier bones. Don't get me wrong--some folks base their entire careers on size/shape relationships, but I find size patently uninteresting (I'm sorry Stephen Jay Gould please don't hate me). Therefore I prefer to control for size statistically and get it the heck out of my way. Fortunately, size standardizations are pretty easy to apply with a few well-written Excel functions [19]. LPT to any budding researchers out there: make Excel bow to you. Sure, statistical platforms get all the glory, but Excel is an underappreciated beast when it comes to stuff like this--plus it gives you the option to autostandardize new data, which will save future you a lot of time [20]. I’ve honestly gotten more mileage out of Excel than I have R (a popular statistical platform).

All told, the entire PhD timeline was about 9 years (~3.5 years to make it through comps and secure project funding cobbled together from various sources, 4 years for data collection and preparation, ~1.5 years to do the analysis and write the dissertation). Of course the dissertation is just one of the responsibilities that grad students have, so I worked on it alongside the many other hats that grad students wear [21]. Did I learn a lot from the experience? Sure--the thousands of hours spent mindlessly executing the same task gave me plenty of time for self-reflection. Would I recommend it to others? Nah--go do yoga or meditate instead.


[1] Naivete is essential for success in academia, as is being incredibly stubborn, unreasonably persistent, and mildly masochistic. Jury is still out on whether you actually need to be smart (speaking for myself).

[2] To the robots trying to steal our jobs, I say “BRING IT!”

[3] I’m stereotyping here, but I suspect the typical 19th-century marriage was not exactly favorable for most queer, over-educated women.

[4] I apologize to all of the trees who died as a result of my research.

[5] Ahahahahahhhaaahhaah haahaahhaa hahahaaahahaha hahahahahahhaa <deep breath> hahahahahahahahhahaaahaha

[6] Fun side note: One author from each of these papers would go onto become my mentor at various points along the PhD process. Brigitte Holt became by PhD advisor, John Relethford became a PhD committee member, and Noreen von Cramon-Taubadel, after years of kindly acting as an involuntary mentor, is now my collaborator, grant sponsor, and boss BHEML where I work. Sometimes life works out pretty OK.

[7] Actual quote: “To date, most quantitative genetic analyses have focused on human craniodental variation, presumably because of the long-held emphasis on cranial variation for understanding human evolution. However, analyses of postcranial variation should play a bigger role in the development of this research agenda. Historically, variation in the postcranium has been presumed primarily to reflect long-term adaptation across species and/or the actions of environmentally or behaviorally driven phenotypic plasticity. However, the extent to which aspects of postcranial variation are the result of neutral evolutionary forces has not been addressed formally”

[8] Personal bias. But it’s totally true.


[10] Want stronger bones? Forget the milk and go lift weights or start couch-to-5k.

[11] I know what some of you are thinking. “Why didn’t you just get a grant to pay for it?” I did, beehatches. Multiple, in fact. But the reality is that when it comes to grad-level grants, even the big ones (NSF, Wenner Gren, etc.) have caps that usually aren’t high enough to afford expensive equipment, let alone expensive equipment + costs associated with data collection (materials, travel, lodging, bench fees, etc.). And don't forget you'll get no salary from them, so you'll be doing all that data collection pro bono.

[12] See footnote [1] again

[13] I chose the latter. That probably explains why I don’t have many European friends despite being there for many months of travel/data collection and my undeniable charisma [13b].

[13b] Not really. I probably fall somewhere between “roadkill” and “territorial North American badger” on the charisma scale.

[14] Never underestimate the power of a good dad joke

[15] Given they already make reality tv shows about extreme couponing, bigfoot hunting, and countless ways to exploit the tired “unique-but-wholesome family” theme, I figure it’s only a matter of time before my packing abilities will be properly recognized on cable television. Perhaps then I’ll be able to retire with some small measure of fame and enough money to support my kitten army.

[16] The Excel file of my dissertation data is so large that it still takes 45 seconds to open it on my brand new 2018 gaming computer. This thing can process entire 3D scans in seconds but struggles with my dissertation dataload. SUPERPROUD.

[17] Yesssssssssssssssssssssssssss!

[18] Booooooooooooooooooooooo

[19] Until two years later when you submit the work for publication and Reviewer 2 invariably complains about the fact you used X size control methods instead of his size control methods (or the literally infinite size controls developed by other researchers) and demands you do the whole darn analysis over again.

[20] Time is currency in academia. The more of it you can squirrel away using computers to do the work for you, the more opportunities you have to take on dry administrative work that is equal parts soul-sucking, unproductive, and unrewarding, but which makes you a departmental darling. And that + visibility, not research + teaching, is how you succeed in academia, friends. Trust me, I learned through this through the Introvert's School of Hard Knocks.

[21] My hats included: teaching over 20 courses at two colleges, working on several publications, presenting at conferences, doing outreach, continuing grad coursework in Developmental and Evolutionary Biology, Biomechanics, and Biological Anthropology, an RA-ship in Comparative Primatology, mentoring multiple undergrads, working on several side projects (like the mouse lemur project), and sitting on numerous departmental committees. I mention this not to blow my own horn (toot toot) but to demonstrate that grad students aren’t just students, they’re instructors, researchers, grant writers, research assistants, outreach and service instructors, and lab coordinators. Odds are by the time students finish their undergraduate years, they’ll have taken more courses taught by grads and contingent faculty (postdocs, adjuncts) than actual tenured or tenure-track professors. Such folks are essential personnel in the machinery of higher education. /soapbox (for now)

85 views0 comments

Recent Posts

See All