Eight Little Piggies (14 page)

Dubois continued to publish at high volume during this entire time. He retained posts as professor at Amsterdam and curator of the Teylers Museum in Haarlem. His numerous papers are matters of public record, readily available in most good technical libraries of natural history. He devoted his work during these years to a series of studies on the distribution and evolution of brain size in mammals. He wrote far more on this subject than he ever published on
Pithecanthropus
. Moreover, he did not pursue these studies as a sidelight or a fresh start after a Trinil disaster. He began his work on mammalian brains as part of a deliberate program to understand the evolution of human intelligence. In particular, he sought to develop a quantitative method for interpreting the brain size of
Pithecanthropus
and its role in human evolution. The brain work was tapestry, not escape.

(I must confess to a particular and personal stake in this issue. At the beginning of my career, I published several papers on the evolution of brain size. I pursued the same basic mathematical strategy that Dubois had pioneered, beginning in 1897—plotting by power functions. I stand with a tiny handful of scientists who first encountered Dubois as an initiator in the study of brain size by power functions, not as the discoverer of
Pithecanthropus
. My strong interest in the history of science also led me to track down the relationship of this work to Java Man and particularly to the giant gibbon legend.)

Dubois began by confronting the problem that had faced scientists since Cuvier made the key observation just after the French revolution: In some meaningful sense, and not from the mere vanity of our cosmic arrogance, humans are the brainiest creatures on earth. But how shall we measure our superiority? Absolute brain size will not do, for whales and elephants beat us by virtue of their larger bodies, and the consequent need for larger brains to coordinate such bulk. But relative brain size (ratio of brain to body weight) will not work either because brain size increases more slowly than body size along the mammalian spectrum from mouse to elephant—and a shrew, by this false criterion, would be the paragon of mammalian intellectual achievement.

Dubois understood, as others had before him, that the solution to this problem must lie in quantifying the normal or expected relationship between brain weight and body weight in mammals—the so-called mouse-to-elephant curve. We know that brain weight increases more slowly than body weight, but by what amount and in what relationship? Dubois’s predecessors had tried ordinary arithmetic plotting, with little success. In 1891, Otto Snell found the basic solution that we still accept and use today: Relationships between brain and body weight are best described by power functions, yielding a straight line when the logarithms of brain and body are plotted against each other. Snell proposed an exponent of two-thirds for the power function—meaning that brains increase roughly two-thirds as fast as bodies along the mouse-to-elephant curve.

Once a standard mouse-to-elephant curve has been established, the right criterion for assessing human superiority (or the status of any particular species among mammals as a whole) becomes clear: Compare the actual brain weight of a species in question with the expected brain weight for an average mammal of the same body weight on the mouse-to-elephant curve. By this proper criterion, no mammal exceeds the positive deviation of
Homo sapiens
.

Snell had established the basis for a solution in 1891, but Dubois’s classic paper of 1897 provided the first extensive data. By comparing seven pairs of closely related mammalian species differing substantially in body size, Dubois recalculated Snell’s exponent at five-ninths—that is, Dubois argued that brains increase five-ninths as fast as bodies along the mouse-to-elephant curve. (Dubois was quite wrong in this conclusion, and his error set the basis for his later problems as we shall see. No single number can capture the range of variation among mammalian rates; the evolution of the brain is not a problem in point masses and frictionless surfaces from a Physics 1 laboratory exercise. Moreover, Snell’s two-thirds generally works better and more often than Dubois’s five-ninths). Using five-ninths as a standard value, Dubois then compared the brains of “intelligent” species with expected brain sizes at their own body weights, thereby establishing a scale of cranial capacity that rightly placed
Homo sapiens
at the pinnacle.

Now Dubois did not begin these studies for idle curiosity or for solace or novelty after anger at criticism of
Pithecanthropus
. He wanted to know the standard relationship of brain and body in order to calculate the intermediary status of Java Man between modern apes and humans. In 1935, he reminisced:

At the Fourth International Anthropological Congress, meeting in Cambridge, England, in 1898, Dubois used the formula from his studies on brain weight to judge the status of
Pithecanthropus
. At a cranial capacity of 855 cm
³
inferred from the skullcap (Dubois later revised this figure upward to about 900 cm
³
), an average modern ape would weight 230 kg and an average modern human but 19 kg. Since
Pithecanthropus
, as reconstructed from the femur, clearly stood between these figures in weight, the Trinil hominid must have carried a brain intermediate in size between modern apes and human. Dubois concluded in 1898: “From these considerations it follows that
Pithecanthropus erectus
is an intermediate form between man and the apes…a most venerable ape-man, representing a stage in our phylogeny.”

Two years later, Dubois locked up the bones, but he never stopped his work on brain size. He amassed a large collection of brain casts, most made by himself, at the Teylers Museum, and he continued to publish substantial papers with important data. At the same time, he developed an evolutionary perspective based on his brain work and hostile to Darwinism.

Dubois began with two assumptions, long common in continental traditions of evolutionary thought, but foreign to English procedures. Dubois held, first of all, that sequences of species with increasingly positive deviations from the mouse-to-elephant curve indicated an intrinsic push towards higher intelligence in the evolution of mammals—an inherent drive that could not be explained by something so base as Darwinian natural selection and adaptation to changing environments. Dubois wrote in 1928:

As a second anti-Darwinian conviction, Dubois followed his countryman and former mentor Hugo DeVries in arguing that evolution occurs by sudden leaps, not gradual transformation. “Again and again,” Dubois writes in striking metaphor, “we find pillars of the expected bridges, never arches.” Applying this assumption to the human brain, Dubois held in 1935: “There was a leap from the anthropomorphous [ape] level to the pithecanthropus level, not the gradual slow ascent presumed by the darwinistic hypothesis.”

Dubois made his fateful error in applying these assumptions to his data on brain sizes. He scanned his tables of deviations from the mouse-to-elephant curve, and managed to delude himself into believing that he could arrange these values into a few discrete groups differing by factors of two. In other words, Dubois thought that he could identify a group of species with a common level of minimum brain size. From this foundation, brains increased by sudden doubling, with the next group at twice the basal brain weight, the next at four times, the next at eight times, et cetera. Starting from the top, and setting the human maximum at one, Dubois claimed standard values of one-quarter for great apes, one-eighth for most carnivores and hoofed herbivores, one-sixteenth for Leporidae (rabbits), one-thirty-second for Muricidae (mice), and, at the bottom, one-sixty-fourth for Soricidae (shrews).

Dubois then took his final step and proposed a neurological and embryological basis for enlargement of the brain by leaps. Neurons are formed in the embryo and do not divide after birth; most mammals enter the world with their full complement of brain cells. Each division of neurons should double the weight of the brain (to grasp Dubois’s theory, we must bypass the obvious, and valid, objection that all neurons do not divide at once on the beat of a metronome). Thus, evolution progresses by inserting one more division into the program of embryological development—one extra doubling towards the human pinnacle. Dubois wrote in 1928: “One segmentation more or less determines the degree of development in the central organ of the animal life, the extent of the outer and inner world of the animal.”

By estimating the adult complement of neurons, Dubois even felt that he could specify the number of divisions from an initial cell. Shrews had undergone twenty-seven divisions. Counting up the chart of doublings, modern humans stand at the apex of intrinsically driven brilliance with thirty-three divisions.

Alert readers will have detected a gap in Dubois’ series and my arguments. Humans rank as one, apes as one-quarter; humans have thirty-three divisions, apes thirty-one. What happened to one-half (and to thirty-two divisions) in this supposedly unbroken sequence of perfection by doubling? Clearly, the creature with half a human brain and thirty-two divisions must be the holy grail of human paleontology—our direct ancestor, the true “missing link” in a world of evolution by sudden leaps.

And now, having presented Dubois’s tapestry, I may finally correct the last and most insidious claim of the standard legend—the dénouement of the morality play as Dubois, in aged despondency and defeat, redesignates his once-proud ancestor as nothing but a giant gibbon. With the tapestry in place, Dubois’s argument inverts to sensibility. Dubois used the proportions of a gibbon to give
Pithecanthropus
a brain at exactly half our level, thereby rendering his man of Java, the pride of his career, as the direct ancestor of all modern humans. He argued about gibbons to exalt
Pithecanthropus
, not to demote the greatest discovery of his life.

The argument is beautifully clear, adamantine even (however wrong in retrospect), once you understand its basis in Dubois’s work on brain size—the five-ninths scaling law of the mouse-to-elephant curve, evolution towards perfection by doubling, and the missing value of one-half human size and thirty-two doublings. Dubois desperately wanted
Pithecanthropus
as a direct ancestor under his evolutionary view. But the brain of Java Man ranked with embarrassing bulk at some 900 cm
³
, or two-thirds human volume. The Trinil femur does not differ in size from our own and suggests a modern human body weight for
Pithecanthropus
in a reconstruction based on human proportions.

Dubois’s hopes seemed stymied. A human creature of our weight, but with two-thirds our brain size, could not be an ancestor in a world of necessary evolution by sudden doubling.
Pithecanthropus
would have to settle for reduced status as a sterile side-branch of the human trunk. But then Dubois thought of a brilliant way out. His initial reconstruction of the femur as essentially human had always been challenged by many scientists. Rudolf Virchow, the great German pathologist, had argued in 1895 that the femur looked more like a gibbon’s than a human’s.

Now suppose that Virchow had been right after all? Then
Pithecanthropus
would have to be reconstructed with the proportions of a gibbon—particularly, with longer arms and a greatly expanded chest and upper body. Such a giant gibbon would weigh, by Dubois’s calculation, just over 100 kg. Backing down the five-ninths curve to an average human weight of some 65 kg—the proper point of comparison with the brain of modern humans—a 900 cm
³
brain on a 100 kg creature becomes, you guessed it, a brain of exactly half our size at appropriate human weight.
Pithecanthropus
had been deftly rescued from the ultimate limbo of an evolutionary dead end, and raised again to the most exalted status of our direct ancestor. Dubois wrote triumphantly in 1932:

In other words, Dubois never said that
Pithecanthropus
was a gibbon (and therefore the lumbering, almost comical dead end of the legend); rather, he reconstructed Java Man with the proportions of a gibbon in order to inflate the body weight and transform his beloved creature into a direct human ancestor—its highest possible status—under his curious theory of evolution.

Eugène Dubois is no hero in my book, if only because I share the spirit of his unorthodoxies, but disagree so strongly with his version, and regard his supporting arguments as so weakly construed and so willfully blind to opposing evidence (the dogmatist within is always worse than the enemy without). Nonetheless, I step forward in his cause because he has been so badly served by careless reading and neglect of his primary work (theories of brain size) in favor of his moment of fame (the discovery of
Pithecanthropus
). As a bitter consequence, Dubois’s ingenious attempt to retain
Pithecanthropus
as a direct human ancestor has been widely misread in a precisely opposite manner as an ultimate surrender, almost comical in its transmogrification of a human forebear into a giant gibbon. With the patch of
Pithecanthropus
alone, the legend of the giant gibbon easily takes root; you need the tapestry of Dubois’s evolutionary theories about brain size, the integrating vision of his intellectual life, to grasp his true intent.