for parallelism by filing until the requisite degree of accuracy is obtained. The parallelism cannot be made altogether perfect, and, as a matter of fact, in some cases the error may amount to as much as one-tenth of a micron or more.

Fig. 71 represents a perspective view of the same thing.FIG. 71 In this figure the intermediate standard rests on a carriage by means of which it may be moved as necessary for the purpose of comparing it with the whole meter. In making this comparison the surfaces must be parallel to the mirror which serves as a reference plane in the interferometer. The parallelism in this case must be of the same order of accuracy as that between the surfaces themselves. The adjustment is made by the screws at the rear, one of which turns the whole standard about a vertical axis and the other about a horizontal one.

In determining the number of waves in the meter, the first operation is to find the number of whole waves in this intermediate standard. It can readily be conceived that the counting of something like 300,000 waves would be no small matter; in fact, a little calculation would show that, if we counted two per second, it would take over forty hours to make the count. Probably a number of methods will suggest themselves of making such a process of counting automatic, Indeed, several experiments have been made, and with some promise of success; but the possibility