_The Nature of the Movements actually caused by Improvements._--This upward movement cannot go on as smoothly and with as little disturbance as that which is caused by the increase of capital. Whenever a greater gain is made at one point than is made at another, an influence is set working which, of itself, tends to send labor from the one point to the other. The slowness with which the change of method proceeds affords the time that is necessary for the protection of labor in the first-named group, since little movement takes place before the effects of improvements made in the second group begin to be felt. If in 1906 an improvement is made which, in the course of five years, would cause some labor to move from the subgroup _A_''' to the subgroup _B_''', and in 1907 a corresponding improvement is made in the latter industry, the equilibrium is restored before enough disturbance has taken place to require any absolute reduction of labor in _A'''_. The facts are (1) that new laborers as they enter the field are drawn more to the upper subgroups than to the lower ones,--to the _A'''_ and the _B'''_ rather than to the _A_ and the _B_ of the two series,--and that in moving upward they are drawn at first more strongly toward _B'''_ and later more strongly toward _A'''_. This is the nearly constant fact in industry and is the grand resultant of all the forces we have described--an upward flow that is continuous but does not follow strictly vertical lines. As young men--the sons of workers in _A_, _B_, _C_, and _D_, who might otherwise have remained in their fathers' occupation--move to the subgroups that stand higher in the several series, they first go in larger number toward _B'''_ than toward _A'''_, and later in larger number toward _A'''_. There is a wavy movement toward the right and then toward the left in the steady flow of labor from the groups that create the raw material to those that impart to these materials the form utilities which they need to fit them for service. An actual lessening of the number of workers in an entire group in consequence of an improvement in the method of production is practically unknown, and even a positive lessening of the number in a subgroup is exceedingly rare.