HUMANE TREATMENT OF ANIMALS USED IN RESEARCH 289 
true is that, in our entire survey of the scientific literature, we have not found 
to exist any theoretical basis for finding out what constitutes, in a statisician’s 
sense, a significant sample for purposes of planning or evaluating an experiment 
on any living animal. 
The implications of this apparently prosaic fact are hair raising. It means 
that the results of experiments on animals are of an entirely different, and much 
lower, order of accuracy from the results of other sorts of experiments. The 
difference can be compared to putting money in Government bonds, as against 
gambling it at the races. It is, in fact, worse than that. At the races, we are 
at least quoted odds against a given horse, a rough probability value. But in the 
animal experiment, where no one knows, no one has discovered, whether a sig- 
nificant sample will be used, or what constitutes a significant sample, the proba- 
bility value of any results obtained does not even exist. It is not defined. The 
experimenter does not know, literally, the degree of uncertainty involved in 
assigning the degree of uncertainty of his results. It is not just a case of the 
odds being so many to one against his being sure. He does not know, and can- 
not find out, what the odds are. He is a man betting in the dark, against un- 
known odds, by some homemade rule of thumb. It is not surprising, therefore, 
that the AMA questions whether much of value, in proportion to the cost, can 
come from his work. But his expenditures consist of dollars just as real, and 
just as valuable, as those that go into atomic submarines or radar warning 
nets. 
CHECKUP ON VIVISECTION GRANTS NEEDED 
Let us take an actual case, to make this point concrete. For example, in the 
experiment to find out what factors influence a monkey to care for its mother, 
the ultimate purpose must be to find out something about the motivations or 
behavior of human beings, if the experiment is to have any utility for us. Hence, 
the chain of reasoning underlying the experiment must run : 
(1) What is true of certain monkeys here in this laboratory is true of all 
monkeys. 
(2) What is true of all monkeys is, to some extent, true of all mammals, 
for monkeys are mammals. 
(3) What is true of mammals in general is true of men, for men are 
mammals. 
Now, right at step (1), this reasoning hits a snag, for the question, “How many 
monkeys must be tested here in this laboratory before we can say, with reason- 
able certainty, that the results are likely to be true of any monkey outside this 
laboratory?” has no answer, so far as the present scientific literature is con- 
cerned. Much less is the answer defined to the question of how many monkeys 
must be tested, with what uniformity of result, before the probability can be 
ascertained that the results will be true of mammals in general, or of men in 
particular. 
If a physicist finds that samples of supercooled boron have certain electrical 
properties, he is justified in publishing his results in terms of boron in general, 
or possibly even in terms of the cryogenic properties of certain groups of ele- 
ments. But the only valid information our monkey researcher can possibly have, 
by the very standards of science itself, refers only to specific monkeys in his 
laboratory, and not even to those as they exist now, but only as they existed 
when the experiments were performed. This is no mere verbal objection, no 
empty technicality. It has to do with the same sort of practical problem as the 
question of the investment of money in blue-chip securities, as against a wildcat 
uranium mining stock. Again, statistically based inference is the only guide we 
have, and the key to reliable use of such inference is a certain minimum amount 
of information, of experience, of standards to go by. 
Now, these facts are true of research on living animals, as they are true of no 
other field even loosely termed scientific. The results of animal experimentation 
are of an entirely different order of accuracy from those of the body of scientific 
findings — a lower order. (Of course, this is not true of the results of work in 
microbiology or biochemistry, which are not faced with the same problem, and 
in which progress has been steady and fruitful.) The animal experiments have 
not, and cannot have, the same order of reliability, or the same value from the 
point of view of prediction, as orthodox scientific studies. It is in the light 
of this indisputable difference that I venture to suggest to you that, quite apart 
from a possible investigation of all types of research appropriations by your 
committee, which you mention in your letter, some sort of permanent check 
and balance might justifiably be set on the appropriation of tax money for such 
animal research projects. 
