396 



PROCEEDINGS OF THE AMERICAN ACADEMY 



of the discrepancy is due to an error by which the interval between 

 the minima of October 3 and 6 is about five minutes too small. This 

 will affect all the minima preceding or all of those following it. The 

 number of decimal places employed by Dr. Schmidt has been retained, 

 although the accuracy of the observations does not seem to justify it. 

 Since we do not know the period within some tenths of a minute, it 

 seems scarcely advisable to carry the result to thousandths of a minute 

 in the formula used to represent it. 



Formula (C) is proposed as that which best satisfies all the obser- 

 vations, if we admit that the period is invariable. 



Formula (D) is that which best represents the later measures ob- 

 tained at the Harvard College Observatory. 



TABLE XV. — Comparison of Ephemerides. 



An examination of the residuals from Dr. Schmidt's formula shows 

 that this ephemeris alone satisfies all of his measures, if we admit his 

 second reduction of his observations in August. It, however, entirely 

 fails to represent the later determinations. The deviations exceed two 

 hours, both in the Harvard College measures and in those of Mr. Knott 

 When this ephemeris was published, tliese observations had not been 

 made, and of course such a deviation could not have been foreseen. It 

 is, however, remarkable tliat Dr. Schmidt should not have noticed the 

 hiro-e discordance in the minimum observed by M. Glasenapp. This ob- 

 servation has especial value as a test of any ephemeris, since it is much 

 earlier than any other measures. Dr. Schmidt's ephemeris would give 

 a minimum at ll"" 32™, Moscow mean time, which is at once seen to 

 be in error on inspecting Table XHI. or the original publication of 

 M. Glasenapp. In fact, the reduction shows a correction to the time 

 of minininm of over an hour and a half. 



