Methods of taking frequency-response curves of loud speakers have tended to become prematurely standardized within commercial organizations, and inadequately standardized in the industry as a whole. Characteristics taken by different experimenters are not comparable. Each may have a good defense to make of his system, and it is the writer's plea that such discussion should be “A free for all, and not a private affair.” More general discussion would promote education if not standardization. A critical attitude on the part of all is desirable, and a demand that published curves be accompanied by information as to method of test. Two valuable contributions on the subject have so far been published, in which the result obtained by several methods of test are compared. This paper makes additional contribution to the experimental evidence, suggests some other methods of measurement, and criticises the several methods from a theoretical standpoint. One of the difficulties arises from the fact that it is desired to show a single curve whose flatness is an indication of the general merit or fidelity of the loud speaker, whereas the differences in directive properties as well as in total sound output make it impossible to describe the properties of a speaker in a single curve. Only in as far as it is possible to specify the exact conditions under which the loud speaker will be used, can such a single curve have significance.