Page:Philosophical Review Volume 2.djvu/702

This page has been proofread, but needs to be validated.
688
THE PHILOSOPHICAL REVIEW.
[Vol. II.

method we can know nothing without the calculation of some one of the characteristic variations; yet the same experimenter remarks that it was hardly worth while to calculate the mean variations, from which we can draw only one conclusion, that it was hardly worth while to present the results at all. Another psychologist rises superior to the charge of not possessing the faintest idea of accuracy by declaring that psychological phenomena are not measurable quantities, that our measurements are physical, etc., not knowing that the science of measurements has stringent rules for all measurements and not seeing that his plea for carelessness simply denies him the right to make any measurements.

No matter how accurate or inaccurate the measurements may be, the amount of trust to be given to the results will be indicated by a proper treatment of the variations and their differences, that is, so far as chance errors and changing systematic errors may have influenced the work. The sources of constant error must unfortunately be left to the experimenter; it is easily seen how fatal the reputation for carelessness must be. There can be no question that the results obtained by many a poor investigator are actually measurements of some error of apparatus or of method and not of a psychological phenomenon at all. One by one we are getting the psychological conditions under control and reducing the amount of error. That some psychologists choose to declare themselves superior to such slow and careful work and prefer to make startling experiments where little or nothing is known of the method or of the complex mass of phenomena measured, is only too unfortunate.

When measurements are made at all, the experimenter must know just how accurate his apparatus, his methods, and his conditions are to be made and are made. Ignorance of the apparatus, laxity in method, and carelessness in work will be shown in the published records, provided a proper account of the apparatus and methods is given and a proper computation of the results is undertaken. In any case where such data are not given, we cannot accept the results.