OF ARTS AND SCIENCES. 383 



the residuals are smallest in the last column, it seems probable that 

 the value of a grade is not always the same. 



The results of the two methods agree as closely as would be expected, 

 even if no systematic errors increased their discordance. The residuals 

 of the photometric observations indicate a probable error of 0.024 mag- 

 nitudes for each group. Assuming an equal accordance in the observa- 

 vations of Schonfeld, the two methods should differ by 0.04, or by the 

 amount found in Table VI. 



DM. 8V25. 



The variability of this star was detected by M. Ceraski, of Moscow, 

 during the summer of 1880. It was soon shown that it belonged to 

 the Algol class, or that every few days it lost a large portion of it3 

 light for several hours ; the interval in the case of this star is somewhat 

 less than two days and a half. Measurements of its light were made 

 according to the method described above in the case of /3 Persei. The 

 photometer was attached to the 15-inch telescope of the Harvard Col- 

 lege Observatory, since as much light-gathering power as possible was 

 desired, owing to the faintness of the star. The same observers took 

 part in the work, and the observations were made in the same way as 

 with y8 Persei, except that the images were reversed by turning the 

 photometer instead of by moving the prism. This could be done very 

 conveniently by a pinion, which served to rotate the entire tail-piece 

 of the telescope. The prism was therefore set once for all, and the 

 images reversed and separated by any desired amount with great nicety 

 by turning a milled head. The star DM. 81°26, which is estimated 

 in the Durchmusterung to be of the 9.5 magnitude, and is nearly 

 north at a distance of 5', was used for comparison. DM. 81°30 would 

 have been better on account of its greater brightness, but its distance 

 of 8' is so great that both images could not be easily brought together. 

 The large angle of the prisms and their distance from the object-glass 

 rendered the light-pencils divergent. At first this gave much trouble, 

 but it was remedied by placing the images always in the same part of 

 the field. Two cardboard points visible against the background of the 

 sky secured this condition. The great northern declination reduced 

 the errors of the driving clock to about one sixth of what they would 

 be for an equatorial star. 



The first measures to determine the full brightness of the variable 

 in terras of that of the comparison star were made on February 6, 

 1881. On the following evening the variable attained its minimum at 

 about half-past eleven. Forty sets or four hundred and eighty set- 

 tings wero obtained between a quarter past six and half past ten. The 



