The Return of the Face

Physiognomy is a discarded 19th-century pseudoscience. Why can’t we stop practicing it?

Adrian Daub | Longreads | October 2018 | 16 minutes (4,170 words)


Physiognomy — the attempt to interpret a person’s character by means of their face — was one of those things that educated 19th-century Europeans knew wasn’t supposed to work. In his 1806 work The Phenomenology of Spirit, philosopher G.W.F. Hegel devoted a lengthy, indecipherable chapter to explain why physiognomy, and its cousin phrenology, had to be hokum. But even if Europeans knew they shouldn’t put stock in physiognomy, they found it incredibly difficult to resist the impulse.

To some extent this remains true today. During the Obama years, many of us were sensitive to representations of the new president, knowing full well that the way faces are read and analyzed could easily encode very old and deeply embedded racist ideas. Then Trump was elected. In a heartbeat, we were back to reading his face, playing with his face, and displaying it next to animal faces. Where does this temptation come from?

A tweet by writer Moira Weigel recently drew my attention to two covers of TIME Magazine, published about 25 years apart. One, from 1993, purports to show “The New Face of America,” which turns out to be a young woman’s face “created by a computer from a mix of several races.” It suggests that America might breed its way out of a long history of racial exploitation if “we” all intermarry enough to make race no longer legible.

The second TIME cover Weigel highlighted perfectly blends Donald Trump’s face and Vladimir Putin’s — an image that was a bit of a rip-off of a 2017 cover by the German news magazine Der Spiegel. The idea that Trump’s true face may be someone else’s has been in the ether since the beginning of his Presidency. And now it, rather than the attractive computer-generated woman, turned out to be “The New Face of America.”

The gap between these two composite faces “seem[s] to capture everything about the last quarter century of liberalism in America,” Weigel says. And indeed, there is a deep link between liberalism and physiognomy. Physiognomy gave concrete shape to liberalism’s dark secret: the sense that reasoning and discourse are only part of how society and politics function, and that just looking someone in the face can be more revealing than listening to what they say.

The two magazine covers express two very different ideas, but rely on a similar notion of what can or cannot be read in a face. Physiognomy dates back to ancient times, but the attempt to systematize and use technology to harness it hails from the 19th century. That age designed an entire arsenal of clamps, holders, cameras, projection screens, and photographic techniques that allowed faces to become readable in a whole new way — laying the foundation for the facial-recognition mania of our own day.

* * *

The story of modern physiognomy begins in the German-speaking world. Johann Caspar Lavater (1741-1801), a Swiss theologian, published his Physiognomic Fragments in four volumes between 1775 and 1778. His books quickly became popular throughout Western Europe, and soon began to influence how people literally pictured themselves. In Central Europe, for example, they led to an obsession with silhouette pictures, usually created by cutting a person’s profile out of black cardboard. It provided an affordable means of having your portrait done, but Lavater gravitated to it because it promised to make faces measurable. My grandparents still had these made for my mother and her sister in the 1950s: each of the two girls in profile, their noses and foreheads lovingly detailed. Their profiles show pigtails tied with identical ribbons that neither woman is confident she ever actually wore, but it’s clear that their facial features are the point of the pictures. They hang above my grandparents’ piano to this day.

Lavater’s apparatus for taking silhouettes (public domain, via Wikimedia Commons)

Physiognomy quickly integrated itself into the bourgeois domestic sphere: It told the well-to-do family about itself. To think of your family in physiognomic terms meant to treat your kin like aristocrats; farmers and workers were frequently featured in physiognomic collections, but it’s highly doubtful they ever consumed them in large numbers. As much as it focused on the faces of others, every book on physiognomy was above all an invitation to anxious self-examination.

Physiognomy usually (though by no means always) abstracted from what a face was doing to what it looked like anatomically. It searched for ridges and protrusions, for pronounced features and underdeveloped ones. As a result, it tended to produce conveniently static design pieces that fit nicely in the décor of a salon or a sitting room: death masks, portraits, attractive coffee table books, all nicely sanitized and turned perfectly private. What a face had done or said out in the world had disappeared in favor of obsessively rendered folds, full lips, and pointy chins.

As much as it focused on the faces of others, every book on physiognomy was above all an invitation to anxious self-examination.

Physiognomic thinking has remained so persistent because physiognomy was always a practice as well as a theory. While it didn’t take a Hegel to point out the theory was exceptionally weak, the visual practices associated with it wormed their way into people’s everyday perceptions. They fascinated criminologists and crime writers, and influenced prominent authors in their descriptions and visual artists in their portraiture. They guided the development of early photography and the popularization of Darwinism. Look at Émile Zola’s physical descriptions, August Strindberg’s characters, the conventions of early portrait photography, or the poses in which politicians and monarchs wanted to be represented, and you’ll find them emphasizing features that physiognomy had taught people to find interesting. Kaiser Wilhelm II, for instance, tightly controlled any photographic representation of himself and enthusiastically distributed carte-de-visite versions of his photographs to anyone who would have them.

* * *

These days we have become a lot more allergic to distinctive facial features. The beautiful face — from fashion models to snapchat filters — is defined by conformity rather than uniqueness. The 19th century, anticipating the way in which the mass would norm faces and expressions, worried that the craggy, unadorned, pre-photographic face was going extinct. At a historic moment at which every face on earth may soon be subject to the algorithms of facial-recognition software, a moment at which AI can harness vast amounts of data to reduce every face to the same factors, their fear may well be realized.

Physiognomy didn’t prove to be dangerous because it worked, but because people believed it worked.

It’s not surprising that during the 19th century, when faith in progress and science were extremely firm, a lot of quack theories thrived alongside the foundations of knowledge we still rely on today, sometimes even purveyed by the same people. Like many bad ideas you can’t quite let go of, the notion that a pronounced brow ridge (for example) indicated animality reared its head in whatever modish guise the scientific consensus of the time would almost, but not quite support.

Kickstart your weekend reading by getting the week’s best Longreads delivered to your inbox every Friday afternoon.

Sign up

How did it come to seem so plausible? Physiognomy was rarely taught in universities or tested by determined amateurs in their labs. Instead, its mode of evidence boiled down to dissemination: teach people how to make certain associations and let confirmation bias take care of the rest. It worked so long as people wanted it to work. In some ways our current debates about facial-recognition software and its use in surveillance aren’t that different. There’s a fair amount of evidence that the technology is nowhere near where it would need to be to realize our Orwellian nightmares. China’s Face++ software requires a passerby to stand still for a few seconds to confirm an ID, not the most likely event in the process of pursuing a criminal. Amazon’s facial-recognition software recently matched 28 Members of Congress to digitized mugshots very much not of those 28 Members of Congress. But our belief in the technology’s capabilities is powerful on its own.

There are plenty of other ethical issues around surveillance technology; the discipline of physiognomy shines a spotlight on a somewhat underappreciated one: Physiognomy didn’t prove to be dangerous because it worked, but because people believed it worked. What gives us the creeps looking back at 19th-century techniques of facial analysis isn’t how well they interpreted broad-backed noses, but the legitimacy they conferred on the attempt to do so. Facial recognition raises the same question: What are the ethical and political stakes in thinking that our faces can be read in certain ways?

* * *

Outside of the arts, the way the 19th century looked at faces had become far more quantitative. Lavater’s rapturous descriptions of his subjects’ features had the tone of devotional literature. Many physiognomists who came after him instead wanted to be scientists — they drew on what we would today call biometric data and harnessed the new methodologies of statistics and demography (the word “biometric” was coined by one of these practitioners, Karl Pearson, in 1893). They quantified, mapped, and tabulated. Franz Joseph Gall (1758-1828) founded the pseudoscience of phrenology, while Carl Gustav Carus (1789-1869) wrote The Symbolism of the Human Form (1853), where he argued that physiological differences explained “the different abilities” of different races. Men like Alphonse Bertillon (1853-1914) and Cesare Lombroso (1835-1909) looked at faces as part of a criminal anthropology that wanted to infer criminality from physical features.

Unlike today’s facial-recognition software, which links the features of your face retroactively to an entry in a database, their interest was predictive: provide the physiognomist with a face he has never encountered before and let him tell you what this person may do or have done. Lombroso would reduce his subjects to a single defining criminal type (“the prostitute,” “the alcoholic”), and then look not for the traces of the behavior, but for the traits that had predicted it. Lavater went searching for “genius” in the faces of his contemporaries. His successors were far less well-intentioned in what they searched for.

In societies that increasingly understood themselves as egalitarian, and where birth was nominally no longer supposed to determine your station, physiognomy could legitimate difference: just looking at this man’s face, can’t you tell why he is the boss and you are not? What physiognomy saw in your face (and to a lesser extent in your cranium and in your hands) were natural factors, not the scars and blemishes left there by hard work, drink, or spending time outdoors. It didn’t try to discern your history, or any history for that matter. Physiognomy couldn’t stop looking at faces, but was purposefully blind to the social dimensions of the human body.

When it did detect those dimensions, it tended to declare them natural. The individual factors physiognomists analyzed in skulls and faces varied from practitioner to practitioner, and also changed over time. But most often they mapped their objects onto a matrix that juxtaposed strength and weakness, crudity and refinement, closeness to nature and “humanity,” defined as distance from animality. In each case, any trait deviating too far from the median in either direction was either primitive or sickly. They were as scared of overrefinement (factors indicating a modern, effete, nervous disposition) as they were of what they termed atavism (factors welling up from human prehistory).

In societies that increasingly understood themselves as egalitarian, and where birth was nominally no longer supposed to determine your station, physiognomy could legitimate difference: just looking at this man’s face, can’t you tell why he is the boss and you are not?

Tangled up in these polarities were profound anxieties about race and gender; it’s hard not to get the sense that physiognomy scooped up every awful idea the 19th century had to offer. In his book About Face, the literary historian Richard Gray argued that we can trace a continuity from Lavater to Auschwitz. Others have pointed out that the reason physiognomy shaped people’s thinking so strongly was that it didn’t always come in its most virulent, racist guise — a lot of it was well-intentioned, a little nostalgic, and ultimately pretty harmless.

Lavater had intended his book to aid with “knowing and loving man” — he was interested in genius, character, and empathy, and wanted to understand outstanding individuals better by means of their faces. By the late 19th century, physiognomy had moved on to faces that could identify people as types, and usually negative ones. Lavater considered his insights scientific, but his method, if you could call it that, amounted to describing every detail of a face in ecstatic, run-on paragraphs. People like Lombroso, by contrast, tried to reduce and abstract — they wanted physiognomy to deal with large groups of people and to tell a quick story of the criminal rather than their crime.

And why wouldn’t they? Theirs was a world of the mass, of the metropolis, of the face in the crowd. How else would you harness its chaos than by tables and numbers? In Fritz Lang’s 1931 film M, there’s a terrific scene where an army of criminologists try to create the profile of the murderer who has been killing young children around Berlin. At one point a graphologist interprets the killer’s writings and, somewhat hilariously, concludes that the man may have mental health issues. As he prattles on, Lang cuts to the actual killer (played by the amazing Peter Lorre) examining his face in the mirror. It’s as though the killer could hear his diagnosis and were trying to locate his madness, his “type,” in his own face.

Screen capture from Fritz Lang’s M (via Film School Rejects)

Whether Lorre’s character finds what he’s looking for is unclear. But Lang had himself found Lorre only by rejecting physiognomy. He had seen him in a play in Berlin and liked him: “my idea was to cast the murderer aside from what Lombroso has said what a murderer is: big eyebrows, big shoulders, you know, the famous Lombroso picture of a murderer.” Lang picked Lorre precisely because he didn’t fit that picture — whatever it is the child murderer seeks in his reflection, or the police seek in his handwriting, they couldn’t find it there.

* * *

It is shocking today just how widely physiognomic thinking circulated. Collections of photographs and books purporting to teach physiognomic analysis were anything but niche publications — they were staples of the library and the coffee table. They asked readers of a certain class and milieu to study others and themselves in a specific way, and technology more than met them halfway. It was only because photographs and photobooks could circulate cheaply that physiognomic thinking could penetrate deep into everyday life.

Photography was also crucial in another way: it made faces reproducible, measurable, and comparable to hitherto unprecedented degrees. The polymath Sir Francis Galton (1822-1911) used a technique he called composite photography to arrive at an average, “normed” face for a large group of people, such as a family. One of the first projects Galton undertook? Identifying a “Jewish type” by means of composite photography. Galton’s technique seemed to mobilize the latest technological advances to capture what Thomas Hardy called “the family face” in his poem “Heredity”:

I am the family face;

Flesh perishes, I live on,

Projecting trait and trace

Through time to times anon.

Somewhat paradoxically, modern technology made visible what was unmodern in faces. Holding up the individual to a photographic record of their ancestor was entirely different from looking up at an obviously idealized ancestral portrait on the mantelpiece. What people noticed and became anxious about wasn’t how much they’d changed from their ancestors, but how much remained the same. Whether they were looking at photos of their own family or of indigenous people in a faraway colony, the camera gave Europeans an entirely new vantage point for judging themselves and other people.

A reproduction of Francis Galton’s “Jewish Type” composite photo. (Image via Wikimedia Commons, CC BY 4.0)

Photography didn’t just change the way people looked at other people; it may have even changed the way people looked. The philosopher Georges Bataille once made the point that the subjects of early photography have a maladroit kind of openness that human beings seem to have lost as they got better at mugging for a camera. Bataille’s suggestion is that human beings normed their own faces and expressions once they understood that they could, and inevitably would, be photographed. It’s impossible to say whether Bataille was right or wrong. But people who 10 years ago had no idea what a selfie was now think about their angles in ways that suggest a profound transformation in our sense of our own faces.

Physiognomists had a similar worry as Bataille: they were sure that there existed an authentic face of a region, of a people, or of a profession, and that intermixing, modernization, and photographic documentation were slowly making it go extinct. In order to capture that face before it disappeared thanks to all the photos, they… went out and took tons of photos. Some of their attempts ominously point forward to the Nazis’ use of physiognomy. The essayist Rudolf Kassner (1873-1959) wrote in 1922 that the “distinctness” and “richness” of “a society, a race, a Volk, and age” depend on its ability to sustain “types.” “To strip a Volk of this potential to form types is to turn it into a mass, to emasculate it.”

People who 10 years ago had no idea what a selfie was now think about their angles in ways that suggest a profound transformation in our sense of our own faces.

The Weimar Republic, the period of German history sandwiched by World War I and Hitler’s rise to power, was a mass society with exceptionally quick turnover in public figures. Leaders were assassinated or resigned, and governments succeeded one another with alarming speed. Instead of one Kaiser for decades at a time, the public found itself confronted with a new, changing gallery of leaders. To people like Kassner, the anonymity of the democratic electorate found its mirror in the anonymity of the people it elected. Many of those who thought like him yearned for a return of the Kaiser, and when one face rose to the top of the heap in 1933, they welcomed him.

In hindsight it’s not hard to see where this kind of thing was leading. Even before 1933, a lot of physiognomic thought got mixed up with “race science” and anti-Semitic propaganda. To give just one example: When the Nazis decided to mount an exhibition to warn the populace about “Degenerate Music” in 1938, they faced a dilemma. In their “Degenerate Art” exhibition the previous year, they had just brought together a bunch of (stolen) modernist paintings and put sneering commentary next to them. Were they going to hire orchestras to perform entire pieces by composers they actually wanted to erase? Their solution was to display unflattering pictures of composers like Schoenberg and Krenek next to some of their musical manuscripts. Look at these faces, look at this handwriting, the exhibition said: clearly these people are degenerate. You don’t have to listen to their music to know that.

* * *

In 1930, the novelist Joseph Roth expressed his suspicion that it was precisely Hitler’s unremarkable face that made it attractive to his followers, commenting on “Hitler’s physiognomy, which has anticipated all the faces of all his voters, and in which each and every one of his voters can see themselves reflected as though in a mirror.” Hitler had the kind of normed face that photomontage could create. If it wasn’t particularly distinctive, interesting, or attractive, that just showed how average it was.

There’s something to Roth’s idea: a leader’s face can embody aspiration, but also the things we secretly know about ourselves. Half of the United States looks at the face of Donald Trump and cannot believe that it is supposed to represent them. And half of the United States seems to look at this sallow, unhealthy, and time-ravaged frame, see the panicked flailing of the tiny-fingered hands, and recognize some version of themselves: Out of their depth, failing upward and ever upward on a warm cushion of unearned privilege.

Where some saw in Hitler some version of themselves and were attracted to it, others rejected the Führer’s claims to leadership precisely because his face was so ordinary, so un-leaderlike. Friedrich Percy Reck-Malleczewen (1884-1945) was an aristocratic and conservative publicist who, precisely for those reasons, came to oppose the Nazis. During the Third Reich he kept a diary, later published as Diary of a Man in Despair, which contained endless diatribes about the new rulers. None among the Nazi leadership escaped his ire, but he reserved particular vitriol for Hitler, of whom he paints a devastating physiognomic portrait: “A jellylike, slag-gray face,” he recalls one sighting of the Führer outside Munich. “A moonface into which two melancholy jet-black eyes had been set like raisins.”

Many of us know how it feels to stare into the face of a political leader and feel nothing but disgust for each and every feature of it. Far fewer of us are familiar with the feeling of looking at a face and thinking ‘yes, this is it.’

It is not an accident that physiognomic thinking usually survives today in negative judgments, above all in caricature, and especially when it comes to political leaders. But the 19th and early-20th centuries had a habit of idolizing political leaders in extremely physical terms that would nevertheless strike us today as kind of abstract. Vladimir Putin’s perennial shirtlessness amuses American audiences, but we get what Putin is trying to project. In the 19th century, it would have been his chin, his fingers, or his profile that would have solicited approving comments.

Amid the sheer volume of Reck-Malleczewen’s ire and hatred for Hitler’s face, it’s easy to miss that he also had a sense of what the face of a legitimate ruler of Germany would have looked like. Less moon-like, less jelly-like, more expressive. Many of us know how it feels to stare into the face of a political leader and feel nothing but disgust for each and every feature of it. Far fewer of us are familiar with the feeling of looking at a face and thinking “yes, this is it.”

Hitler spent his first few years on the political stage not showing his face. At least some of that was calculation: why would people trained to rhapsodize over the bushy, craggy faces of the Hohenzollern Kaisers suddenly give their vote to a guy whose very face struck the philosopher Theodor W. Adorno as “a composite of King Kong and the suburban barber”? “An apparition with a face like this would have been disobeyed as soon as its mouth spoke an order,” Reck-Malleczewen fumed in his diary. “And not merely by the higher officials in the ministry: no, by the doorman, by the cleaning women!”

When in the late 1920s Hitler’s face went from being nowhere to being everywhere, his enemies unleashed a predictable torrent of venomous diagnoses. But there were as many Germans meditating on the ugliness, meanness, or simple nullity of his visage as there were Germans contorting themselves into finding something in Hitler’s features. In 1933, the philosopher Karl Jaspers went to visit his friend Martin Heidegger, who had not yet fully embraced the Nazi state. Jaspers pleaded with Heidegger, pointing to Hitler’s vapidity and cruelty, but Heidegger responded, “just look at his wonderful hands.”

* * *

That the same face could inspire derisiveness and adulation at the same time might well be a feature rather than a bug. What better way to show which tribe you belonged to than by finding something amazing in a thoroughly unamazing face, or to vituperate endlessly against a visage that to others seemed unremarkable — or even pretty?

In 1831, the cartoonist Charles Philipon had to defend himself against charges of lèse-majesté in court, and defended himself by claiming that “anything can resemble the king,” which he demonstrated using a pear. Philipon was convicted anyway, but the pear became a symbol for a king who had promised liberalization and instead cracked down on free expression. And the way in which it came to represent, but was not allowed to represent, the king indicated that after two French Revolutions things hadn’t much improved. The caricaturist Honoré Daumier in 1834 used the pear in a cartoon juxtaposing past, present, and future: all three are Louis Philippe’s faces on a pear (each more surly-looking than the next), an endless dreidel of the king’s moods. For Daumier, the very stability and continuity that royalists liked about the restored French monarchy constituted an indictment of the same.

Honoré Daumier, Les Poires, 1831. (Image via Wikimedia Commons, public domain)

Physiognomy always had something tribal about it: what you see in someone’s face depends on who you are. Invented by white men, it found beauty, seriousness, and “humanity” in faces wherever they resembled those of white men. Daumier, for instance, satirized the French state and society before the 1848 revolution. Many of his satires exoticize the fashion and politics of his age by making them non-European: he gives French politicians, aristocrats, and society ladies the mien and attire of people from China, Africa, and the Middle East.

In other words, Daumier attaches to French high society the faces of the people their country was busy adding to its empire; even when they were engaged in European self-critique, physiognomists fell back on the aesthetics of imperialist societies. The general mania for faces brought together the photographs that well-to-do Europeans took of themselves and those they took to document the people they were seeking to dominate and exploit.

This kind of tribalism was precisely what men like Lavater meant to overcome. Many aspects of physiognomy mark it as an anti-liberal project, but in one respect it is profoundly liberal. If we were to find one single, universally valid way to interpret faces, we could remove the suspicion that what we recognize in faces we like is simply ourselves, our people, our tribe. We could look at faces without false narcissism and false exoticism. This surely was the intention behind TIME’s “New Face of America” cover back in 1993. And the Trump-Putin composite may well constitute its bitter punchline.

* * *

Adrian Daub is professor of Comparative Literature and German Studies at Stanford University. He is the author of four books on German thought and culture in the nineteenth century, as well as (with Charles Kronengold) The James Bond Songs: Pop Anthems of Late Capitalism. He tweets @adriandaub.

* * *

Editor: Ben Huberman
Fact-checker: Ethan Chiel
Illustrator: Katie Kosma