vidual. His first inference is that by these words the individuals in question intended to convey certain specific ideas, and that in the acts they were guided by certain specific purposes; but he may have wholly misunderstood them both. His second inference is that people wishing to convey such ideas or acting for such purposes must have such and such characteristic conceptions and feelings; but again he may be mistaken. His third inference is that what is characteristic of this score or more of individuals whom he happens to have met is characteristic also of the seventy-six millions whom he has not. Once more he may be wrong. Yet he calls it all observation.

So with scientists. Their inferences are more careful; but still they often use the word Observation to include them. Astronomers, for example, may speak of the observed course of such and such a comet, when they have only observed a few of its positions and have calculated all the rest. Indeed in the strictest sense of the word they can hardly be said to have observed even a single position. They have perceived a speck of light at a certain apparent position in the field of a telescope; at about the same time they pressed an electric key connected with a clock; and they have afterwards read off certain figures from various parts of the telescope and its attachments. That is all. The position of the comet even at one moment is obtained from these data and others like them by elaborate calculations. Thus the word 'observation' is used in science as well as in common life in a very loose sense that is likely to deceive because it seems to imply a closer contact with immediate experience than it really does.

Even when we realize all this and do not regard anything as an Observation except a direct perception we are not yet free from inference. Perceptions are not bare sensations. They are sensations interpreted, and the interpretation, however rapid and involuntary, is an inference, and may be wrong. I see my brother across the road, but when I cross to speak to him it turns out to be some one else; I hear some one say Be honorable', but he really said 'It's in an enve

lope', and so on. These are what psychologists call Illusions. Then again we have Hallucinations, when the interpreted sensation itself arises wholly from within. I hear my name called, see a flash of light, or feel a drop of rain or the crawling of ants, when there is nothing there at all. But because I say 'There is a sound, a light, or an insect, instead of merely saying 'I have' such and such a feeling, or that such and such a feeling exists, I am drawing an inference once more, and once more it is or may be wrong. The only way to avoid all chance of Illusions and Hallucinations is not to

interpret any sensation. But a live person cannot do this; and if he could and did, he would die.

Thus the paradox: induction sets out to base its inferences upon the observation of facts, but the observation is itself a matter of inference. What then shall we do: reduce the element of inference in our observations to the very minimum mentally possible, or let it reach a maximum? What we

really do in most cases is to infer without scruple until something makes us suspect that we have been deceived or that we are dealing with a class of facts in which we are likely to be deceived; and then, if it is not too late, we turn back and examine the phenomena more carefully and critically. No one ever thinks of distrusting his 'senses' so much when he is watching a farmer or a carpenter at work as when he is watching a conjurer or a 'medium'. In the one case we include a great many spontaneous inferences in our servations' and say that we saw him do thus and so; in the other case we only say that he seemed to. What we do naturally in this respect is perfectly logical; for no apparent perception can be tested except by the surrounding conditions as we know them.

'ob

All the difference between the absurd credulity or incredulity of ignorant peasants and the reasonable judgment of the educated depends upon the different extent of their knowledge about phenomena like those in question or the wider world in which they occur.

Credulity.

A scientist

never thinks of doubting the existence of other men, and thus when he 'sees' a colleague in the room with him he believes that the colleague is there. An unsophisticated peasant never thinks of doubting the existence of ghosts, and so when he sees' a ghost in the room with him he believes that it is there. Logically the scientist and the peasant are in precisely the same position, and unless we are willing to say that the scientist should not be so credulous when he believes that he sees something, we have no right to say it of the peasant. Incredulity in general is no better than credulity. The scientist would probably distrust his observation' of the ghost, but he does not distrust his observation in general. Indeed he distrusts his 'observation' of the ghost only because he trusts his other observations and the inferences he has drawn from them enough to believe that ghosts probably do not exist. Thus in observation, as in everything else, general faith precedes and has a logical right to precede specific doubt. The doubt does not come sponta

neously or for its own sake, but it is forced upon us by our faith in our other observations and the larger system of things which they seem to have revealed.

If the field of our observations is a new one and we cannot tell where it is that the spontaneous inferences which we naturally include in our 'observations' are most likely to be wrong, the only thing to do is to go ahead bravely yet cautiously, placing provisional confidence in our observations everywhere, yet always ready to turn back and re-examine any point more carefully. The scientist cannot avoid blunders; for he sees and hears as other men do, and he draws all their spontaneous inferences; but unlike them he knows how much of what he seems to see and hear is really inference, and how likely it therefore is that some of his observations' are erroneous. Consequently he generally takes pains to verify his observations before he announces them, he states them modestly when he does, he expects others to verify them for themselves before accepting them, and he is

willing to be corrected when he has made a mistake. In all of which respects he is very different from most of those who have not received his training.

Since what we 'observe' and the faith we put in it depend upon what we already believe, it is perfectly evident that a wrong belief to begin with will lead to wrong ''observations', so that when we are once started Errors

cumulative.

on the wrong track we keep going further and further, and generally find 'sufficient proofs' for our false convictions.

Trifles light as air

Are to the jealous confirmations strong

As proofs of holy writ.

We pile proof upon proof until at last we stumble across some fact so obvious that we cannot ignore or distort it, or until we discover that the general conception of things that led to so many bad 'observations' is inconsistent with some other general conception just as well established'. Then comes the doubt, the true testing of the observations', and the better general standpoint. The only point in favor of the scientist as contrasted with other people is that he is on the watch for such inconsistencies, and therefore corrects his blunders of theory and observation sooner.

This danger of observing' what we expect to observe and ignoring what we do not is inevitable. Often we wish

to submit a fact to an unprejudiced observer. If we mean by an unprejudiced' observer one who has abso

6

observers.

lutely no convictions that can possibly affect his Unprejudiced observations, then the only unprejudiced observers in the world are newly born babies who do not even believe that there are things and people. But their freedom from prejudice makes it impossible for them to observe' anything at all. When we demand the testimony of an 'unprejudiced observer' the most that we can really wish is that the observer in question shall have no more personal interest in one side of the question at issue than in the other, that he

shall have conceived of both sides as distinctly as possible, and that he shall then have made his observations dispassionately, calmly, and deliberately for the sake of deciding as fairly as possible between the two sides. We do not ask that his mind shall be free from all preliminary convictions, but only from those special convictions the truth of which is disputed by one side or the other. He is not, and cannot be, without prejudice in general, but only without prejudice on the questions involved in this particular dispute. For the rest, we must expect him to take something for granted. Hence if the question changes after the observations are made and now the dispute turns upon some point that our unprejudiced witness never seriously questioned, his testimony is no longer of any special value. And so, in general, an observation made for the sake of settling one question has very little value for the settling of another. When the question is changed the observations should be repeated.

Errors of observation are divided by logicians into those of Mal-Observation, where we perceive things wrongly, as has just been explained, and those of Non-Observation, where we fail to perceive or take account of certain important facts at all. Here are some examples

Two classes of errors.

of the latter.

"Most of the books do not give us a psychology, but rather a eulogy, of animals. They have all been about animal intelligence, never about animal stupidity. . . . Human folk are as a matter of fact eager to find intelligence in animals. They like to. And when the animal observed is a pet belonging to them or to their friends, or when the story is one that has been told as a story to entertain, further complications are introduced. Nor is this all. Besides commonly misstating what facts they report, they report only such facts as show the animal at his best. Dogs get lost

hundreds of times and no one ever notices it or sends an account of it to a scientific magazine. But let one find his way from Brooklyn to Yonkers and the fact immediately