For thermometers that failed to pass the preliminary test, for reasons specified in the first four sections of this table, a rebate is allowed at the pro rata rate corresponding to the number of thermometers submitted for test.

FISCHER.

used without regard to the corrections, and in cases where the accurate temperature is desired the corrections may be applied.

Of the clinical thermometers tested during the year ending July 1, 1904, 88.2 per cent satisfied all the requirements and received certificates, S.5 per cent were rejected because their errors exceeded the allowable error, 1.6 per cent because the mercury index was too difficult to throw back, and 1.5 per cent for other causes, such as defects in construction, moisture, or air bubbles in mercury, etc.; 0.67 per cent were broken when received and 0.50 per cent were broken during test. In conclusion, the authors wish to acknowledge their obligations to Dr. S. W. Stratton, Director of the Bureau of Standards, for many valuable suggestions in the design of the apparatus; to Dr. Chree, of the Kew Observatory, where clinical thermometers have been tested for many years, and to Dr. Wiebe and Dr. Hebe, of the PhysikalischTechnische Reichsanstalt, where such work is done on the largest scale, who gave the authors every facility to inspect the apparatus and methods in use at these institutions; to Mr. L. G. Hoxton and Mr. H. C. Dickinson, who were actively associated with the authors in the work. We are also pleased to acknowledge our indebtedness to the manufacturers for placing at our disposal for investigation considerable numbers of clinical thermometers, and also for much valuable advice and many suggestions. All the testing apparatus described was constructed in the instrument shop of the Bureau.