In 2008, an entrepreneur named Seph Skerritt was frustrated with the way he shopped for clothes. Then a student at MIT’s Sloan School of Management, he chafed at the time wasted while trying on garments in stores. Often, he thought, you settled on an ill-fitting item just to get the drudgery over with. While on an internship in Asia, Skerritt had encountered the effortless magic of having a tailor custom-fit your shirt. Why not improve on that concept, he wondered, with an online service that fitted your shirts by asking you questions, and then mailed you the garments?
He christened his company Proper Cloth. Naysayers told him that when customers input their measurements, they often made mistakes—the idea wouldn’t scale. But Skerritt thought that guessing, even if one’s guesses were occasionally off, was still preferable to the chaos and disappointment experienced in a physical store.
So he set about developing an algorithm that could customize your shirt without needing a tape measure. As a check against errors in customers’ reported measurements, he thought up a list of basic questions—height, weight, and so on—that could serve as indicators of shirt size. Then, using these questions, he made shirts for 30 guys who worked at the New York City tech incubator hosting his startup, called Dogpatch Labs.
When the volunteers tried on their shirts, Skerritt quickly saw what worked and what didn’t. Asking about waist size was insufficient, for example, because it gave no indication of the size of one’s midsection. So Skerritt added a question about how far one’s belly protruded. Other questions were too confusing, like one about how T-shirts fit around your chest and shoulders. Those queries were omitted.
He noticed an odd pattern. In that first batch of 30, the shirts fit best on testers who were Caucasians. They seemed to fit worse, in a predictable way, on people who weren’t Caucasian. All subjects of one ancestry—Asian, say—seemed to require the same general alterations. Skerritt noted the anomaly and added a question on what he called “ethnicity”: Asian, Black, Caucasian, Hispanic, or “I’m not sure.” The question, Skerritt says, has proven invaluable to sizing his customers’ shirts.
There’s no denying the satisfaction of a smartly tailored shirt. But with this one question, the once mundane world of dress shirts is now dabbling in a kind of racial profiling. Are we ready to dredge up centuries of racial strife, simply for a perfect fit?
I bet you have two warring opinions of this web site’s “ethnicity” question. One is that we humans have a long history of buying clothes without explicitly considering our ancestry, so this innovation sounds, if not racist, at least racially inappropriate. The other is that, well, maybe our body types do differ by race—and just accepting this reality frees us from having to wrestle with the Caucasian body proportions that dominate most clothing design. So here’s my question: With the “ethnicity” question, is this entrepreneur courageously addressing the proposition that we’re different according to our ancestry, and propelling us toward a post-racial future? Or is he pretending to be scientific as a marketing gimmick, while actually enforcing false, outdated and possibly dangerous ideas about race?
Past attempts to target clothing to an ethnicity have sparked some controversy. In 2007, for example, Nike launched a line of sneakers decorated with colorful geometric patterns and arrowhead designs for Native Americans, called Air Native N7. Native Americans had wider fore-feet, Nike claimed, and thus needed wider shoes.
But from the moment he heard about the shoes, Alan Goodman, a biological anthropologist at Hampshire College in Amherst, Mass., suspected that Nike’s science was weak.
Nike says it measured 224 Native Americans’ feet around the country before concluding that Native feet were wider in the front. Who was measured, Goodman wonders, how old were they, and what was their condition? Native Americans have double the risk of diabetes compared to the national average. Was Nike really selling colorful, “native” sneakers for swollen diabetic feet? Or as one online commentator put it, “If this isn’t an example of corporate manipulation of race, I don’t know what is.” (I reached out to Nike for a response, but never heard back.)
When I asked about Proper Cloth’s “ethnicity” question, Goodman had this to say: “Calling groups white or black is a pre-Darwinian view of biology that does not fit the facts of human variation.” Other anthropologists I spoke to also roundly denounced the question. Race, they say, is a social construct.
Here’s what they mean: if you were to travel across the Eurasian continent from Portugal to Japan, say, there would be no river, forest or discrete boundary where people suddenly started looking “Asian” (or “European” if you traveled from east to west). Instead, changes in physical appearance would occur so gradually and imperceptibly that you probably wouldn’t notice. You’d only be aware that, once in Japan, people definitely looked different than in Portugal.
The assertion is that the traits we associate with “race”—hair color and texture, skin pigmentation, epicanthic eye folds and whatever else—actually occur in gradients throughout the human family, not clean breaks. So much variation occurs within what we call “races” that to ascribe any particular traits to a certain race is, anthropologists say, misguided.
Before we slam down the gavel on Proper Cloth, though, a quick story: I’m lanky, with long arms, long legs and a decidedly short torso. I can be the same height as someone standing but when I sit, they tower over me.
Some years back, I found myself in Seoul, South Korea, perusing dress jackets in a department store. As I tried them on, I discovered that all the sleeves were several inches too short for my lanky arms. I am not (to my knowledge) Asian. Apparently, in Seoul, my body type was an extreme outlier.
So let’s look at the situation as Proper Cloth might: If that Korean department store had had a “western” section full of jackets sized more like those I might find in New York City, would that really be so bad?
When I related the tale to Goodman and asked if what I’d perceived was real—that I had an unusual body for Seoul—he said, “Yeah, that’s real, man. But Koreans are not a race. And you’re not a race. You’re an individual variant who happens to have long arms.”
He was objecting to how we talked about the differences in our bodies, not that the differences existed. He was saying that casting these distinctions as “racial” was incorrect (and calling these labels “ethnicities,” as Proper Cloth does, is no less problematic). It’s also potentially dangerous.
Anthropologists are keenly aware of this peril, because it comes from their own field. It was early “anthropologists” (those are air-quotes) who first described a “natural” caste system of humans that featured certain whites on top. That pseudoscience was then used to justify slavery, imperialism, colonialism, Nazi atrocities and more—to repeatedly excuse the brutal exploitation and even extermination of one group of people by another.
The other, more concrete problem was this:
Sure, it’s easy to find generalizations in the anthropology literature about how people of West African descent have longer limbs compared to their trunks than Caucasians, and Asians smaller limb-to-trunk ratios. And there’s even a plausible scientific reason why all animals, including humans, might be stockier in colder climes and lankier in warmer ones: to better conserve and dissipate heat, respectively.
But where exactly did those limb and torso measurements come from?
Let’s say you’re a character on a CSI show, and one day a colleague lugs a sack of human bones into your lab, dumps them on an exam table and asks: “Was this person black or white?” This is actually someone’s job, it turns out. Among other tasks, forensic anthropologists must determine the “race” of human remains from little more than bones. And you can find statistics claiming they’re right more than 80 percent of the time, implying that, even without skin and flesh, the shape of our bones says something about our race.
Here’s the problem, though: When Goodman looked into a few old studies, he discovered the conclusions hadn’t held up to later scrutiny. By his estimation, forensic scientists trying to determine race were actually correct about one-third of the time—no better than chance. They might as well be guessing.
Joe Hefner, a forensic anthropologist at Michigan State University in East Lansing, echoed Goodman’s critique. A lack of scientific rigor has irked him since his early days in the field. You’d ask a forensic anthropologist how they arrived at their conclusions—how they knew a skull was white or black—and you might hear, “because I’m an expert,” he said. The old guard seemed to work off intuition. They couldn’t explain their methodology; their techniques couldn’t be taught.
Even if you had good information on the range of features in the human family—meaning you’d collected many measurements from a random assortment of people, and you’d done so without knowing anyone’s “race” beforehand—you’re still just making an informed guess. That guess, says Hefner, should be properly expressed as a probability. As in: “with this distance between the eyes sockets, and this head width, the subject has a 60 percent chance of being white, and a 40 percent chance of being something else.”
Again, Hefner and Goodman weren’t saying that human variation didn’t exist, or even that it was impossible to gauge. They were arguing that given the ignominious history of race as a concept, the science had to be done carefully and rigorously. And that just hasn’t happened.
So how, they both wondered, could an internet-based shirt company claim to pull it off?
For over a century, the U.S. military has measured incoming recruits’ bodies. In the Civil War, height, weight and BMI were used to determine physical fitness, and to exclude men with diseases, such as tuberculosis. In World War I, these measurements helped determine how much recruits could carry, and how far they could march. After World War II, the importance of knowing recruits’ body shapes and sizes became apparent for other reasons: to properly fit soldiers into the machines that were now central to warfare. In what’s still cited as a notorious example of unfortunate design, some gun turrets for the B-17 bombers were so cramped that only the smallest crew members could sandwich their bodies into them, and in a semi-fetal position at that.
This series of measurements, comprising thousands of recruits, is known as ANSUR—the Army Anthropometry Survey. And it includes race. When I asked, anthropologists tended to pooh-pooh the dataset. Among other objections, they noted that military recruits were not a random sample of Americans. But I encountered a different opinion among industrial designers.
Matthew Reed, a researcher in anthropometry at the University of Michigan in Ann Arbor, flat out called ANSUR the “best-gathered anthro data in the world.” Not only was the dataset accurate and reliable, in his view, it could save lives.
Imagine body armor that doesn’t completely cover your vital organs, or is too long or too short, making it horribly uncomfortable to wear, so you take it off. Imagine if you found yourself in a nerve gas attack and your mask doesn’t fit properly because it was designed for a different kind of face. (One study found that American-made gas masks fit poorly on “Chinese” faces.)
The differences in body type according to race could be striking. Reed sent me a graphic, based on data from the 1988 iteration of ANSUR, illustrating seated height versus standing height for Caucasian and African American men. Most African Americans had a shorter seated height compared to Caucasians of the same overall stature, meaning longer leg bones and shorter torsos. Asians, meanwhile, skewed slightly in the other direction, with taller sitting heights and longer torsos than both Caucasians and African-Americans.
The army doesn’t use this information to individually fit uniforms and gear, Reed explained, but to plan and manage costs. If the army knows that 15 percent of recruits are African American, when it orders, say, 20,000 bullet-proof vests, it will ensure that 15 percent conform to what it believes is their relatively shorter proportions. “That’s really important for the army,” Reed told me. “If you do that wrong, you end up with stuff that you need to store. And you don’t have enough of what you do need.”
Reed points out that race is also important in civilian contexts. Think about your car. Reed designs crash test dummies. If a car is tested only with “Caucasian” dummies, it may not be as safe for Asians or African Americans. Why? Leg length determines how far back you sit from the steering wheel—a major impact point—and your proximity to the airbag. Seated height also affects what you can see. “We don’t want to build a dummy that’s based only on white guys,” Reed said.
The point is not to design “black” or “Asian” dummies. Rather, it’s to ensure that the dummies you use are as diverse, in their body proportions, as you know the American population to be. In short, while some anthropologists argue that we should ditch the old race labels, Reed considers taking race into consideration good design practice.
And here’s what kept nagging at me as I reported: even as science tells us there’s no biological basis for many of the ideas we’ve inherited on race, one can sense a yearning in this soon-to-be majority minority country for acknowledgement that we do actually differ according to our ancestry, and that we shouldn’t all be held to one, in this case mostly northern European, physical standard. This yearning often seems to bubble over in discussions of beauty. Worried that her daughter will hold herself to a white standard of beauty, one black mother writes in The Guardian, “I underestimated the effect of the exclusion of black female bodies from the American beauty pantheon.” Black female role models like Serena Williams and Misty Copeland are thus doubly important as highly visible reflections of black, female beauty. “Williams is a cultural touchstone for everyone,” she writes, “but for young black girls, living in a culture that rarely defines their appearance as beautiful, she is as necessary as air.”
The dream here is to recast the physical differences that underlie what we call race as mere variation, not deviations from an ideal norm. But here’s the conundrum: Is that accomplished best by ignoring race (as some anthropologists argue), or by explicitly acknowledging it, even if racial differences aren’t as real or predictable as we’ve been led to believe, and even if the language we use to talk about them is archaic?
“It is NOT racist to acknowledge that particular ethnicities are more likely to have a particular body type,” writes Natasha Devon, a mixed-race body image expert, in The Independent. “What IS racist is to silently but powerfully expect women of all cultural backgrounds to conform to Caucasian beauty paradigms.”
Maybe not surprisingly, it’s retailers—the people trying to sell you stuff you probably don’t need—who have most wholeheartedly embraced this idea. In the early aughts, fuller-figured, curvaceous mannequins began arriving to store windows, touted as a different ideal of female beauty—one that owed more to J. Lo and Beyoncé than to Kate Moss. These mannequins say nothing about how our bodies differ by ancestry, of course, but they reflect different standards for how we want out bodies to look—standards that some scientists say roughly correspond to race and ethnicity.
Then there’s more practical stuff. A number of eyeglass companies now market “Asia”- or “Asian-fit” frames. The glasses are designed for a less prominent nose bridge and a wider face. Writing at Refinery29, Connie Wang called the moniker “backwards-sounding.” But she also expresses relief to have found glasses that don’t constantly slide off her face, and make her feel like she has, in her words, a “deformed head.”
We could dismiss this as racial profiling, eyeglass edition. But notice the outcome: a better fitting pair of glasses, and a positive shift in how a woman sees herself in relation to an imagined norm.
Here’s the origin of Proper Cloth’s ethnicity question. When Skerritt discovered the race anomaly in his first batch of shirts, he sought out prior research—corroboration that we actually do differ according to ancestry. He came across the military’s research on body type and race, and adjusted his algorithm accordingly. But the algorithm has evolved since those early days. It continues to learn based on patterns gleaned from Proper Cloth’s now 20,000-person strong database. When I first tinkered with the algorithm, I entered my dimensions—5’11” and 170 lbs—and then toggled between races. The “black” me had slightly longer sleeves than the Asian or Caucasian me. The Hispanic me had a wider chest. For some reason, the “black” me also had a longer shirt than other versions of me, which seems to contradict the ANSUR data indicating shorter “black” torsos.
When I asked Skerritt about the results, he said that his algorithm has uncovered some associations that weren’t necessarily predicted by other research. For example, if I put in another weight and height—let’s say 5’4” and 170 lbs—and THEN changed ethnicity, Skerritt told me, the predictions for other parameters, like neck circumference, might change slightly. He couldn’t really explain these anomalies. Nor could I verify that his shirts actually fit better because of the ethnicity question. The company’s data is proprietary.
But what Skerritt was saying, essentially, was that the algorithm was a kind of scientific pioneer, probing the heretofore unexplored community of Men Who Buy Fitted Shirts, and making new discoveries.
When I’d talked to Hefner and Goodman, they’d warned that most anthropometric data sets were unreliable: the data collectors weren’t blinded to their subjects (which could introduce bias), the participants weren’t randomly selected, and the sample sizes were sometimes too small. Yet Skerritt seemed to be fulfilling at least two criteria for good data collection—that you take measurements blinded (in this case, with customers entering their own information)—and then see if patterns show up when you reveal subjects’ race. And he had a large sample size.
What about just four “ethnicity” categories for all of humanity? Skerritt was thinking about adding more options, he admitted. But he also didn’t want the algorithm to become cluttered. “People on the internet are not super patient,” he said. Okay, but didn’t the Hispanic question seem a little—well, simplistic?
All “Hispanic” means is that one’s ancestors come from the remnants of a fantastically large empire, emanating from the Iberian Peninsula and comprising huge swaths of the Americas. But Hispanics can be of African, European, Native American, Middle Eastern, and East Asian descent—or various combinations of these and more. Meaning that, checking Hispanic would tell you nearly nothing about one’s body. Even Matt Reed, who championed using race as a proxy for body shape, thought that question too vague to be useful.
Did Skerritt have any idea which Hispanic population—Puerto Ricans, Mexicans, Colombians—was buying his shirts? He didn’t. And he wasn’t sure it really mattered. Even without that level of detail, the question seemed helpful in sizing shirts, he said.
Here Skerritt echoed something Reed had said.
I’d asked Reed how he justified using a fiction (race) to measure reality (body shape). Reed responded that while, yes, race was a social construct, because people tended to think about themselves in predictable ways in relation to racial labels, race could still yield solid information.
Skerritt made a similar argument in defense of his “Hispanic” label. How could “Hispanic” produce meaningful information about body shape, given how little it indicated about genetic ancestry? For the algorithm to work, Skerritt said, it doesn’t need to accurately track a person’s ancestry—or for that matter, his true belly size. Respondents only need to answer the question to the best of their abilities. Because people tend to make the same mistakes, or tell the same lies, or have the same misconceptions about themselves—because none of us are as original as we like to think—they invariably produced similar responses, yielding the patterns necessary to fit your shirt.
“The algorithm doesn’t care what you really are,” he told me. “It matters what you choose, and what you think you are.”
I was inclined to think that, as a company that needs to profit to stay afloat, Proper Cloth wouldn’t waste resources on frivolous, pointless questions—that the ethnicity question must help, as Skerritt insisted. But to the end, Goodman remained “highly skeptical.” Only direct evidence that the question decreased returned shirts would convince him of its legitimacy.
In a sense, his insistence on seeing for himself was an argument for scientific transparency—something private companies are not by any means obligated to do. But it was also a challenge to how we’ve all been conditioned to think about race. He was essentially saying, Prove that it exists. And if you can’t or won’t prove it, then stop talking about it. Because without proof, the concept of race, as it pertains to variation in the human family, is too dangerous.
Let’s open a conversation about this. I’m curious to hear people’s lived experiences in this arena. Please write a response below if you’ve encountered troubles finding well-fitting clothes potentially due to your ancestry, or share your diatribes for/against the “reality” of race.
