288 HUMANE TREATMENT OF ANIMALS USED IN RESEARCH 
Knowing that you have long and ably stood for the proposition that a dollar 
of our tax money wasted is in effect a dollar contributed to communism, I would 
like to take a little space, and a little of your time, to inform you of something 
we have come to know about animal research projects, simply from the point of 
view of their scientific worth, and to put forward a suggestion which I believe 
might interest you. I do not propose to take up your time by reiterating our 
main arguments, with which I know you are already familiar. What I propose 
to do here is talk about fundamental scientific principles, and about the economic 
principle of a dollar’s worth of value in exchange for a dollar paid out 
A FALSE ANALOGY TO REAL SCIENCE 
I respectfully suggest to you that the real cause of the current difficulty with 
medical research stems from a false analogy between the physical sciences and 
the biosciences. Our Government has, over the years, acquired experience in 
allocating funds effectively and fostering useful research in the former; it was 
only natural that with the rapid rise of the latter (which are still very new), 
the same procedures should have been adopted. But it is my purpose to demon- 
strate here that the present procedures for allocating funds for medical research 
have not yet been adapted to reality or logic, on the basis of pragmatic tests 
which our democratic form of government has always demanded in spending 
the taxpayer’s money. 
I say that there has been a false analogy drawn between the physical sciences 
and the so-called life sciences, to the extent that methods proven in one area 
have been uncritically applied in the other. Let me demonstrate what I mean, 
and at the same time illustrate our reasoning in asserting that live animal ex- 
perimentation is inevitably sloppy science. 
TBUE SCIENCE GIVES WORKABLE RESULTS 
For a physicist or a chemist, there is a sufficient body of experience accumu- 
lated, and a sufficiently tested general theory, to make it safe to assume that one 
atom of, say, copper is just like (for all practical purposes) another atom of 
copper. There is sufficient experimental evidence already accumulated to justify, 
even, the extrapolation of some results gained from experiments on copper to 
applications involving, for example, silver, or in some eases even plutonium, or 
perhaps generally all metals. The laws involved, however, are statistical laws. 
They speak in terms of probabilities, ranging in value from 1.00 (certainty) to 
0.00 (impossibility) as limits. In practice, these limits are, of course, never at- 
tained, even in the most precise experiments. The scientist, always and forever, 
because the reasoning of science is inescapably inductive in nature, must deal 
with probability values. This fact has, through the writings of scientists, 
become familiar to all of us. Almost as familiar to the man in the street is the 
idea that, for a statistical generalization to represent a scientific truth, a suffi- 
cient number of cases must be examined to give validity to the probability 
values. The statisticians and mathematicians have, as you know, worked this 
out quantitatively, and have arrived at definite calculations by which it is pos- 
sible to find out the minimum size for a significant sample, the least number of 
individual cases from which, in given circumstances, it is safe to generalize. 
Naturally, the greater the number of cases tested, up to a point, the safer is 
the inference to be drawn from them. But below a certain number of cases (the 
significance sample), it is not safe or valid to draw any inference. To reason 
from too few cases is to fall into the same error which has given the world such 
superstitions as that about the ill luck derived from a black cat, or breaking a 
mirror. Given certain data, the actual numerical size of the significant sample 
can be computed, in true sciences, before the experiment is conducted. 
EVEN AMA RAISES DOUBT ABOUT EXPERIMENTS 
Now, research animals are infinitely larger than atoms (and infinitely more 
expensive to keep about). They are also infinitely more various. Standard 
strains of mice have been developed, but they are standard only with respect to 
a few very limited parameters. Even the famous fruit flies of the geneticists 
( Drosophila melanogaster) are not perfectly standard. There is no really stand- 
ard animal, no standard experimental dog, or cat, or monkey, or guinea pig. 
Every animal differs from every other. And every animal, naturally, differs ac- 
cording to external conditions, from one day to the next. What is shockingly 
