Permanence of the Rutherfurd Photographic Plates. 215 



two screw corrections belonging to the two numbers whose dif- 

 ference is set down in the column headed " Difference of Means." 



This method of reduction has the advantage of showing ver}^ 

 clearly how much the results have been affected in each individual 

 case by the application of the screw corrections. The line begin- 

 ning " Mean of Differences " gives the mean of the measures of 

 the east and west impressions, together with the mean of the two 

 corresponding screw corrections. This number being expressed 

 in revolutions of the screw needs to be divided b}^ 2 in order to 

 turn it into divisions of the scale. The next line, therefore, con- 

 tains the half of the number in the preceding line, together with 

 the corresponding half of the screw correction. This last has 

 now become only 2 units of the fourth decimal place. The 

 next line contains the division error of the line numbered 34 on 

 the scale. This was the line read in the present measure, and the 

 <Jivision error for it was taken from the table on p. 206. The last 

 line then contains the point on the scale that is exactl}^ opposite 

 this star, according to the present measure. The number is ob- 

 tained by applying the division-error correction and the tiny 

 screw correction to the number marked " ^m." The whole num- 

 ber, 84, is simply the number attached to the line read on the 

 scale. 



It sometimes happens that the same line is not read on the scale 

 for both the East and West impressions. For we alwa3^s read on 

 the next smaller division line of the scale, and if a line of the scale 

 happened to be between the East and West images, the lines read 

 for the two images would not be the same. In such cases it is 

 evidently necessary to take the mean of the lines as well as the 

 mean of the microscope readings. Thus if the line read for the 

 West impression in the above example had been 35 instead of 34, 

 the final line at the end should be 34^ instead of 34. This would be 

 expressed by writing the final number 34.6022 instead of 34. 1022. 



6. The table given below contains the results of all the mea- 

 sures, set down in chronological order, according to the date of 

 measurement. Two lines are devoted to each set of measures like 

 the one just described. All the measures of each star taken on 

 the same day are placed side by side, as it was thought this ar- 

 rangement would make reference to the table more convenient. 

 Of course the measures were not taken in the order as set down, 

 but were made in strict accordance with the principle of repetition 



