o fifty per cent, and thirteen still more than fifty per
cent. But besides the omissions there were only six among the forty
which did not contain positively wrong statements; in twenty-four
papers up to ten per cent of the statements were free inventions, and in
ten answers--that is, in one fourth of the papers--more than ten per
cent of the statements were absolutely false, in spite of the fact that
they all came from scientifically trained observers. Only four persons,
for instance, among forty noticed that the negro had nothing on his
head; the others gave him a derby, or a high hat, and so on. In addition
to this, a red suit, a brown one, a striped one, a coffee-colored
jacket, shirt sleeves, and similar costume were invented for him. He
wore in reality white trousers and a black jacket with a large red
neck-tie. The scientific commission which reported the details of the
inquiry came to the general statement that the majority of the observers
omitted or falsified about half of the processes which occurred
completely in their field of vision. As was to be expected, the judgment
as to the time duration of the act varied between a few seconds and
several minutes.[16]

Another type of cases in which our direct testimony would be valueless
is legerdemain: we think that we actually see rabbits taken out of our
neighbor's hat, or his watch pounded in a mortar and presently shaken
whole and sound out of an empty silk handkerchief; and it is only by
reasoning that we know our eyes have been deceived.

It is obvious, therefore, that to question a man's evidence is not
always to call him a liar; in most cases it is rather to question the
accuracy of his inferences from such part of the facts as he actually
grasped. In science no important observation is accepted until the
experiments have been repeated and checked by other observers. Indeed,
most of the progress of science is due to the repetition of experiments
by observers who notice some critical phenomena which their predecessors
have missed.

With this qualification, that human observation is always fallible,
good direct evidence is on the whole the most convincing evidence that
you can use. If you can establish a fact by the mouths of trustworthy
witnesses, making your readers recognize that these witnesses had good
opportunities of observation and a competent knowledge of the subject,
you will generally establish your point. In case of an accident in a
street car i