Here, however, simple observation stops, and we must have recourse to experiment.
You tie a vein, and you find that the blood accumulates on the side of the ligature opposite the heart. You tie an artery, and you find that the blood accumulates on the side near the heart. Open the chest, and you see the heart contracting with great force. Make openings into its principal cavities, and you will find that all the blood flows out, and no more pressure is exerted on either side of the arterial or venous ligature.
Now all these facts, taken together, constitute the evidence that the blood is propelled by the heart through the arteries, and returns by the veins—that, in short, the blood circulates.
Suppose our experiments and observations have been made on horses, then we group and ticket them into a general proposition, thus:—all horses have a circulation of their blood.
Henceforward a horse is a sort of indication or label, telling us where we shall find a peculiar series of phænomena called the circulation of the blood.
Here is our general proposition then.
How and when are we justified in making our next step—a deduction from it?
Suppose our physiologist, whose experience is limited to horses, meets with a zebra for the first time,—will he suppose that this generalization holds good for zebras also?
That depends very much on his turn of mind. But we will suppose him to be a bold man. He will say, "The zebra is certainly not a horse, but it is very like one,—so like, that it must be the 'ticket' or mark of a blood-circulation also; and, I conclude that the zebra has a circulation."
That is a deduction, a very fair deduction, but by no means to be considered scientifically secure. This last quality in fact can only be given by verification—that is, by making a zebra the subject of all the experiments performed on the horse. Of course, in the present case, the deduction would be confirmed by this process of verification, and the result would be, not merely a positive widening of knowledge, but a fair increase of confidence in the truth of one's generalizations in other cases.