William Howard Taft
The Greatest - or at least Largest - Chief Justice

The Best of Jobs/The Worst of Jobs

William Howard Taft always preferred being Chief Justice of the United States over all his other jobs. Earlier he had been Solicitor General, Justice on the US Court of Appeals, Governor of the Philippines and Cuba, and then the United States Secretary of War - an office that later became the less bellicose sounding Secretary of Defense. He also had some other job. Just can't think what it was.

But it was the job of Chief Justice that William really liked. He sat on the highest bench for almost a decade, from 1921 until shortly before his death in 1930.

Big Vinegar: Truth with the Intent to Deceive

Some rulings of William and his court have relevance for us today but seem a bit odd and amusing when described. One case involved a company that claimed to be making vinegar. More specifically they said they were making apple cider vinegar. But instead of taking apples and squashing them to get the juice (which when left standing eventually becomes vinegar), they were taking dehydrated apples and rehydrating them. That is they mashed the dried apples in water. Then they took the water and made their vinegar.

Now it was true that the product the company produced analyzed chemically as vinegar. It also looked and tasted like vinegar. So what, we ask, was the problem?

Well, the Pure Food and Drug Act had been pushed through by William's friend, Teddy Roosevelt, in 1906. This statute had distinguished an actual product from a substitute. It also established criteria for listing proper ingredients and gave guidelines for proper labeling.

But with the new process, the company was actually using flavored water - from steeping the water in dried apples - and not real apple juice. They also said they used "selected" apples in their process. So courts were asked to rule if the product was really apple cider vinegar.

And eventually the case wound up at the highest court in the land. There the court ruled - in a unanimous opinion - that the product was not really apple cider vinegar as the consumer understood. Real apple cider had to come from apple juice - not water mooshed into dehydrated apples.

Also "dehydrated" apples could not be properly dubbed as "selected apples". The latter designation implied to the typical consumer that the apples had been specially chosen and screened before use - not that they were just dried out apples.

So although the product thus made might be a type of vinegar or perhaps a vinegar substitute, it was not apple cider vinegar. So the product was mislabeled. The point of the case was that although there were no outright falsehoods stated, the wording on the labels was misleading. In other words, truth with the intent to deceive is not allowed.

William's Really Big Case: Who Can Be an American?

You'll note that the decade of the Taft Court was also that of the Roaring Twenties. This was a time of unprecedented prosperity where families in the lower economic strata broached the increasingly affluent middle class. Ironically, the 1920's was also a decade of rising xenophobia, which was at least partly fueled by rising immigration and the increasing activity of the Civil Rights Movement.

Among the cases heard by William and his court were whether it was legal for states to prohibit teaching foreign languages (the court ruled it was not), if primary elections could exclude black voters (the answer was no), if baseball was a business and subject to anti-trust laws (the court ruled it was neither), and whether the United States could deny citizenship to immigrants based on their race. It's this last case that we'll look at.

The United States Constitution states somewhat ungrammatically that Congress has the power to "establish an uniform Rule of Naturalization". Other than that pithy statement, there are no limitations or guidelines. In other words, who and how non-citizens can be granted citizenship is up to Congress and no one else.

The first naturalization law was passed during George Washington's first year of office. And Congress made being a naturalized citizen pretty exclusive. They stated that applicants had to have resided under US jurisdiction for 2 years, and - here's the kicker - they had to be - quote - "free white men" - unquote. A later law passed in 1795 during George's second term said candidates for citizenship also had to be "of good moral character".

As the years rolled on, the residency requirements stretched out. By 1900, you had to have lived in the United States for fourteen years before you could even apply for citizenship. The consequence was that there were more and more "aliens" living in the United States - not because they didn't want to become citizens - but because of the backlog - and the Federal law - they weren't allowed to.

However, some criteria had loosened up. In 1870 the law was modified to allow former slaves to become citizens. But except for this concession, everyone else still had to be of the "white race".

The problem, as we all know, is that no one is really "white". Instead, everyone's skin is brown - it's just a matter of degree. So a case came before William and his court. But they were not asked to rule on whether the laws excluding citizenship based on race were constitutional. Instead they were asked how to define the "white race".

In 1918, Bhagat Singh Thind, a Sikh who was born in Punjab, India, applied for citizenship, and it was granted. But the US Immigration Service reviewed the case. They quickly ruled that Bhagat, who ahd earned a Ph. D. from Berkeley and received an honorable discharge from the US Army after serving in World War I, was not "white". So they revoked his citizenship.

Dr. Thind, though, pointed out that he was a Sikh. And anthropologists had classified Sikhs as "Caucasian". This, Dr. Thind pointed out, was the scientific term for the "white race". And indeed, a number of Indian immigrants had been previously granted citizenship on that basis.

This was not a trivial case. With the rising xenophobia some states had even passed laws that increasingly limited the rights of non-citizens, even if they had been granted permanent residency.

So an adverse ruling for Dr. Thind meant that previously naturalized citizens could actually loose their citizenship. Loss of citizenship would then mean an individual's personal assets - including land - could be seized, even though they and their families had been previously recognized as full citizens. Dr. Thind's case wound it's way through the legal system and ultimately ended in William's rather substantial lap.

The court sidestepped the issue in a very strange way. They would not try to determine Mr. Singh's race by scientific means. Instead they would use the "common man's" definition. To make a long-winded decision short, the court decided that Dr. Thind looked different to your average Joe Blow who considered themselves white. And if he looked different, he wasn't white. So Dr. Thind could not become a US citizen.

The Supreme Court - under William or anyone else - has never reversed this ruling that eligibility for naturalization can be determined by how people look. Instead the courts have abided by the principle that Congress calls the shots. So changes on who can be citizens had to come through legislation.

Gradually things did change, and in 1935, Congress passed a law permitting citizenship for all US Veterans. Then in 1946, a new law specifically allowed immigrants from India to become citizens. Finally in 1967 Lyndon Johnson signed a bill that ended immigration quotas based on national origin and hence of "race". That was also the year that Dr. Thind died in Los Angeles.

And yes, he was a US citizen. He had taken advantage of the 1935 law. He was, after all, an honorably discharged World War I veteran. He also married an American woman, became a lecturer, and wrote ten books.

Defining the Undefinable

As we said, the court ruling - and by the way, William did not write the opinion - seems odd but illustrates the way judges handle sticky problems. They either take common definitions or sometimes don't even bother defining the terms at all, saying it's unnecessary.

But by evading the issue, the Court was admitting what has become increasingly evident. There simply is no way to scientifically determine race even though it remains a crucial characteristic in American law and forensic science.

Most scientists who try it admit defining race is like fighting bees. But why, you ask, if you can't scientifically determine race, why do forensic methods work so well in doing just that?

For instance, we've all seen the news stories where the crime scene detectives determined the bones found in the woods were from a white male. Some news stories even quote forensic scientists saying they can determine race as accurately as they can determine age.

Hm. As accurately as they can determine age.

Now if you watch talking heads on television or read articles in popular magazines, you might hear a forensic anthropologist say things like the examination determined that John Doe died when he was 30 years old plus or minus 3 years. Maybe if they're really uncertain, it'll be something like plus or minus 5 years. But the uncertainty numbers are generally pretty small.

How do they come up with these numbers? Well, training, education, and experience are the usual answers. And we won't quibble with that. Instead we'll run an actual test where we can quantitatively measure how accurate the forensic estimations really are - that is, how much error there is.

OK, we call in three qualified and expert forensic specialists - we'll call them Tom, Dick, and Harriet - and ask them to provide estimates of the ages of 15 skeletons. But these aren't just any skeletons. We know the ages when they died. But we won't tell Tom, Dick, and Harriet.

Tests where we know the answers but the specialists do not are called "round-robins". And round robins are the best way to determine the true reliability of a scientific method. That is, to validate the method. And it is validation tests that allow us to determine how close the experts can get to the real answer and how much they agree with each other. In other words, if the test is any good.

Given how important they are - the outcome of forensic tests can literally be a matter of life or death - published examples of round robins are surprisingly hard to find. After all, few experts like to be put on the spot. But if you dig deep enough you can sometimes find the occasional round-robin article.

And after a long and diligent search we finally found the round robin of Tom, Dick, and Harriet. After their examination of the bones - actually computer scans of the bones - here is what they came up with.

True Age	Age Estimations
	Tom		Dick		Harriet
	Lower	Upper	Lower	Upper	Lower	Upper
54	60	80	49	59	45	65
55	55	75	59.5	74.5	54.5	69.5
40	49	69	47	67	48	58
49	59	69	42	62	45	65
21	26	46	21	41	18	30
28	42	62	30.5	55.5	41.5	56.5
21	19.5	28.5	27	27	20	40
17	22	42	18	28	28	38
28	29	39	30.5	45.5	22.5	37.5
45	35.5	60.5	42	62	48	58
35	28	48	25.5	40.5	38	48
54	59	79	56	66	54	74
57	54	74	65	81	44.5	69.5
34	23.5	48.5	27	47	44	54
57	59	69	39	59	65	65
35	28	48	25.5	40.5	38	48
54	59	79	56	66	54	74
57	54	74	65	81	44.5	69.5
34	23.5	48.5	27	47	44	54
57	59	69	39	59	65	65

Now, to avoid people saying we are being critical of individuals, we want to point out that we are not using data that was taken from any one article. But we have taken great care - and we emphasize this point - that the data we have from Tom, Dick, and Harriet (which are not only pseudonyms but completely made up identities) is indeed data that has the same amount and type of variability that you do see in an actual round robin that was published in the fairly recent literature.

Perhaps it was because they knew they were being tested, Tom, Dick, and Harriet seem to have been cautious. As the table shows, the ages are given as ranges not as individual numbers. Also the ranges seem - at least to us - pretty large, certainly larger than the ranges tossed off by the talking heads we see on television.

So what do we do with this stuff? Well, in this intensely visual age, we plot the data out.

First, that there is a real trend is not to be denied. We clearly see a general upward trend with estimated age vs. true age. In fact, using fancy pants number crunching, statisticians will tell us the confidence that the trend is real is greater than 99%! Such a number is what you read in news articles of how a particular test was - quote - "accurate to 99%" - unquote.

But - and it's a big "but" - that doesn't help us much in determining the age of an individual.

Look at it this way. From the graph we see that if a group of forensic anthropologists were examining a skeleton of someone who died at age 28, the graphs shows us the estimated ages can range anywhere from 20 to 60 years. And if a person was really in their mid-50's, then the experts might tell us the age was 40 - or maybe 50 - no, maybe they were 80!

In other words, that the forensic anthropologists on the average can do better than chance alone in estimating the ages is indeed high - > 99%. But the reliability for them to correctly predict the specific age of an individual is quite low.

And we can use our data to determine the actual variability or uncertainty of an individual predicted age. We get this by a least squares fit from the data. Do that and we get:

Age Estimation (Years) Least Squares Fit
Slope	Intercept	Root Mean Square Error	95 % Confidence Interval
0.92	11.25	10.80	21.18

We see that what data specialists call the root mean square error is about ten years. That means the correct age will be within 10 years of your estimated age only 66 % of the time. Worse, 95 % of the time, the correct age can only reliably fall within 20 years of the estimate age! So if the estimate of the age is, say, 40 years, then we can state with 95 % probability the true age is somewhere between 20 adn 60 years old.

In fact, if we just rely on the estimates, there is also a fairly high systematic error. Note in the table that the intercept is 11 years. That means that on the average Tom, Dick, and Harriet will estimate the ages to be over a decade lower than they really are. So they have a random error of 10 years and a systematic error about twice that large.

Not, we must say, terribly impressive.

As an aside, these tests were run using skeletons in good condition and with known ages. Estimates using bones from archeological sites - or forensic finds - are often fragmentary and in poor condition. Then the estimates can be way off. In one archeological dig, a series of skeletons were subjected to forensic tests by a number of anthropologists. The root mean squared error was 20 years which means the 95 % confidence interval would be plus or minus 40 years! Such an error seem huge but is born out by looking at the individual estimates. In one case, an anthropologist deduced a skeleton was of a 1 year old infant. Another thought the same skeleton was of a 60 year old man!

Clearly such uncertainty makes the estimates of a single skeleton to be of questionable usefulness. Certainly the uncertainty is greater than the - quote - "3 to 5 years" - that the talking heads often throw out. And remember, here we are using forensic methods to determine what is a quantifiable objective property - age - and not what is a cognitive perception - race.

So saying that "race" can be determined as accurately as age is not very reassuring.

All this leads us to a natural question. Have there been any "round robins" for determining race?

Well, not exactly, but there are validation tests to check the accuracy of determining race. For such validation we need to find a "homogeneous" population, make the measurements, and feed the numbers into a computer program that is commonly used by forensic scientists. For instance, if you have some bones from an ancient civilization from Sub-Saharan Africa you expect the "race" to be, well, a race that comes from Sub-Saharan Africa. And you expect the forensic analysis to agree that the population was of a race that lived in that area.

So not too long ago a group of anthropologists did just that. They made measurements of some skulls excavated from a site of the Meriotic Civilization - that is, the African civilization that flourished in the Sudan around 2500 years ago. They took the needed measurements and plugged the numbers into a widely used computer program. Remember: All the skulls were from a single civilization from 2500 years ago in Sub-Saharan Africa.

And what was the computer breakdown of the "races"? Well, it was:

Predicted Races of Meriotic Civilization Burials

Ha? (To quote Shakespeare). What the hey is going on?

Now objections to using this program for such a test are that 1) improvements have been made in computer programs since the article was published and 2) the data used to "calibrate" the program did not have data from the Meriotic civilization. But both statements - although correct - miss the point.

First, it is true that including the Meriotic skull measurements to calibrate the program might have improved the accuracy. But all that means is you have to tell the computer the answer before it can give you the answer! Or to put it in more high-falootin' language, to validate a scientific calculation you have to use known data not used to calibrate the method.

Which is exactly what we did. The Meriotic civilization was from the area of Biblical Cush. The inhabitants were Nubian - that is, they were black but the set of skeletons was not used to calibrate the program. So the Meriotic skeletons can be used as a proper test for validation.

And yet when we did so, modern forensic methods told us that group of black people were made up of more whites than black - and they had some Chinese, Japanese, and Native Americans thrown in as well!

What we're seeing is the problem that has always bedeviled anyone trying to "objectify" race. In other words, how do you translate the way groups of people look into ways that identify their group without looking?

Well, what is usually done is to find groups of people that everyone agrees are of different races. Next we measure their physical characteristics - skull characteristics, bone ratios, and things like blood type.

Then using fancy pants statistical methods, we find that the average measurements between the groups are different - and they differ in a statistically significant manner. What that means is we can be 95 % confident that the average differences will not occur simply by chance. So we think we've found a scientific basis for determining races.

But there's one thing which goobers everything up.

Note that with predicting age, there was a definite trend in the average estimated age with the true age. But the changes in the average estimate changes was less than the "scatter" - the variability - around that age. So we could not identify the age of an individual with any reliability.

And as with determining age, just because we can be 95 % confident that average differences between racial characteristics of population groups are not due to chance, that does not mean we can pick out an individual and be 95 % confident of assigning him to the correct race. We repeat these are not the same thing. Not by a long shot.

Instead variation of characteristics between individuals of the same race are often greater than the variations of the average characteristics between races. That means that individuals will have attributes that fit with more than one race. So you'll always be making mistakes. And as here, you'll make some whoppers.

But how then can you determine race?

What may shock! shock! some people is that the concept of distinct races is a fairly modern idea. Travelers during the early Renaissance - like Marco Polo and Muhammad Ibn Battuta - certainly recognized physical differences in the people they encountered. But because they traveled so slowly they saw a gradually changing continuum. So if you want to classify the members of Kublai Khan's court as a distinct race from the Italians, then you're going to have to start dividing up the continuum. The divisions, though, will be completely arbitrary, and so race ends up being a human construct.

So when did we get the modern idea that we had distinct races? This seems to have arrived with the later era of transoceanic exploration and colonization where the populations were separated by thousands of miles of water. So there were large gaps in the continuum.

Or look at it this way. If you walked from England to Asia and crossed the Bearing Strait and moved down Canada into the Mississippi Valley and then to coastal Virginia you would see a continually changing populations where where all neighboring groups looked essentially alike.

But hop on an ocean going ship and cross the Atlantic and you would see an abrupt change in your hearty Englishman and the aboriginal people of North America. Hey, presto! you've discovered a new race!

Then along came the Restoration and Georgian eras - the 17th and 18th centuries - when natural philosophers began to develop classification systems. And so the notion of continuity was lost and the concept of distinct races gelled, particularly as new boundaries of countries and territories began to be established.

And the new boundaries meant that what started out as a few races kept multiplying, and by the nineteenth century the idea of race had fragmented to absurd proportions. Separate races were designated for the San (Bushmen), Black Africans, Japanese, Chinese, Koreans, Pygmies, Australian Aborigines, Polynesians, Asians, Eskimos, American Indians, Indian Indians, and Melasians. Even different Europeans were classified as different races. People started talking of the "Irish race", the "English race", the "Italian race" and in the ante-bellum South of the United States, there were people who grumped about all the problems they were having with the "Yankee race".

Now you'll notice something very interesting here. For the most part, we see that "race" was really grouping people by linguistic families. And it has turned out that today talking about linguistics families has been far more meaningful and useful for historians than talking about race, even in countries where polyglots abound.

Race, though, remains a human cognitive construct. Or to put it another way, because races are constructed by arbitrarily partitioning a continuum, a race must remain a defined concept. So we can invent races any way we want. Which is what ultimately what William and his court did.

William was, if not the greatest Chief Justice, then certainly the largest Chief Justice. Bouncing about - no wisecrack intended - between 300 and 350 pounds - he had to have special bathtubs installed, although the story he once got stuck in the tub and had to be rescued is of dubious veracity. William also had problems finding clothes that would withstand his variable expanse. When he was Governor of the Philippines, he visited Czar Nicholas in Russia. As he climbed down from the carriage he heard a rending sound. His pants had split down the backside. Mrs. Taft, Helen (Nettie to her friends), asked one of the attendants for a needle and thread and came to William's rescue.

But of course, William did have that other job - the last holder, by the way, to sport a mustache or other facial hair. William held the office only a single term and tried for a second. But he was soundly trounced by the clean-shaven Woodrow Wilson, a man who also left a decidedly mixed legacy - not bad on the international forum, but horrible for Civil Rights.

A big factor in William's loss was because his now former friend, Teddy Roosevelt, dissatisfied at the directions William's more traditional and less flamboyant presidency was taking the Republican party, formed the new Progressive Party, which was popularly dubbed the Bull Moose Party. With the Republican vote split, the Democrats were a shoo-in.

As for William, we can cut him some slack and say he did the best he could. But all in all he has not been afforded great respect. Once Clarence Darrow was invited by a friend to attend a speech by "Big Bill" before a Joint Session of Congress. Clarence sat through the speech in silence. Finally, he leaned over toward his friend who waited eagerly for Clarence's words of wisdom:

"Fat son-of-a-bitch, ain't he?"

References

There doesn't seem to be any really definitive biography of William except for a couple of books published shortly after he died. Instead the few books there are tend to be rather dated or cover specific aspects of his life.

The William Howard Taft Presidency, Lewis Gould, University Press of Kansas, 2009.

Presidential Anecdotes, Paul Boller, Oxford University Press, 1996

Clarence Darrow for the Defense, Irving Stone, Doubleday, Doran, 1941

"Forensic Misclassification of Ancient Nubian Crania: Implications for Assumptions about Human Variation", Frank Williams, Robert Blcher, and and George Armelagos, Current Anthropology, 46 (2), 2005.

"Racial Identification in the Skull and Teeth," Jodi Blumenfeld, Totem: The University of Western Ontario Journal of Anthropology, Vol. 8, Issue 1, 2000.

"Race as a Visual Feature: Using Visual Search and Perceptual Discrimination Tasks to Understand Face Categories and the Cross-Race Recognition Deficit", Daniel Levin, Journal of Experimental Psychology, Vol. 129, No.4, pp. 559-574, 2000.