Page:Popular Science Monthly Volume 6.djvu/414

398 server reads the microscopes of a Transit Circle habitually too large, when he is determining the zenith-distance of a star, it is likewise his habit to read them too large when determining the position of the zenith-point from which zenith-distances are counted; and the resulting quantity is likely to be free from all but accidental errors.

Occasionally there arise cases where these differences (in the same observer) are not eliminated, but multiplied.

In the measurement of a base-line, for example, the various rods are brought into contact under a microscope: if an observer judges these rods to be in contact when they are not, it is evident that his error, originally small, will augment with the number of contacts, and it may become serious.

In the comparison of the national standards of length, undertaken by the English Ordnance Survey, an annoying case of personal difference was found.

These comparisons were made by bringing a movable cross of spider-lines to bisect one of the lines engraved, on the various bars, and it was found that Captain Clarke, R. E., and Quartermaster Steel, R. E., who made the greater number of comparisons, differed in their estimation of a bisection by a constant amount which was annoyingly large: so that "the probable error of the final results is nearly double what might be expected from errors of observations only." This error cannot be eliminated, and it still remains in the published results.

We must constantly bear in mind that the quantities of which we have all along been speaking are extremely small, and that in fact they are masked by accidental errors for inexperienced observers in most cases. Still they exist, and they are among the most curious of phenomena: their careful study would well repay physiologists.

We can never be sure we have eliminated them so long as the human mind or body is a part of the machine by means of which we are comparing or registering events; and, just so long as mind or body is employed, we can be sure that personal differences will not only exist, but that they will vary from day to day. We must use for eliminating personality those values which are the best attainable, and assume these values to be constant over extended periods of time—weeks or months. In astronomy of precision, however, we have other errors to fear much more variable than personal equation, and it is to the elimination of these that attention should be directed. In other branches of research less exact in method, personality becomes of more importance, and an attentive consideration of its effects may be well worth while undertaking.