_Spread._ _Square end._ _Forked end._ _Round end._
one plane, varies varies middle pair curved up, middle pair middle pair middle pair curved down, all all all _Closed._ one plane, outer pair outer pair middle pair different planes, outer pair outer pair all
Which shape brings the most feathers into use in all positions? Which positions bring most feathers into use? We see at once that the rounded end has a decided advantage, that the middle pair of feathers is used in all possible positions, that the pair next outside is the next important, and that the spread tail curving downward at the edges and the closed tail in different planes are the two shapes which give the best support. There is therefore a reason for the rounded end which we said was the rule among the woodpeckers.
Our little experiment is what we call a _deduction_. It shows us what we ought to expect under certain imaginary conditions. But it does not show us what actually exists, so there often comes a time when our deductions are faulty because Nature has done some unexpected thing, as when we found the single exception of the logcock's foot upsetting a fine theory of ours. A deduction must always be compared with facts, and is worth little or nothing if a single fact of the series we are studying is not explained by it. This time all the facts do agree; for I had, before we made our experiment, examined the tails of every species of woodpecker ever found in North America, and there was no exception to the rounded end. I had already drawn my conclusion that this form was better adapted to life on a tree-trunk than the square or the forked tail would be, reasoning by a different process called _induction_. An induction examines many, and, if possible, all the facts before drawing any conclusion; a deduction examines the facts after the conclusion is reached. There is no hard-and-fast line between the two kinds of reasoning, but we may say that a _deduction is reasoning out a guess and an induction is guessing out a reason_. Deductions are easier and quicker; inductions are surer, and in preparing them we often make other discoveries.
The rounded tail is no doubt the best; but we have yet to decide whether the sharper curve is more advantageous than the lesser curve, as we thought probable from our observations. And there is still another deduction from our experiment which we did not make. If in the rounded tail the middle pairs of feathers do most of the work, and if use increases the size and efficiency of a part, which is almost an axiom in science, we should expect to find the middle tail feathers not only strongest in all woodpeckers but also strongest in increasing ratio in the species that use them most. To determine this we must study the use of the tail and the structure and shape of the individual tail feathers.