REPORT OF THE SECRETARY. 71 



2. Imperfect adjustment of the prism-mirror combination for minimum deviation — 



Errors from this cause proved negligible under the assumption of parallel rays. 

 As steps are now being taken to introduce a system of collimation for the 

 beam with a view to improve the definition of the optical system, further 

 discussion for a divergent beam is unnecessary. 



3. Changes in the temperature of the prism — 



No new data on the change of deviation with the temperature have been 

 obtained, but referring to those of a preliminary nature contaiued in your 

 paper on the "Temperature of the Moon/' 1 it is believed that errors from 

 this source in curves taken tinder usual conditions will not exceed 2 seconds 

 of an arc in deviation. With the attainment of accurate data such errors 

 may be corrected. 



II. Sources of error which introduce accidental deflections in the record cur res. 



1. Disturbauces primarily affecting the galvanometer needle including (a) earth 



tremors; (6) magnetic disturbances. 



2. Disturbances secondarily affecting the galvanometer needle through the bolome- 



ter circuit including (a) those causing "drift;" (b) those causing "flutter" 

 in the records. 

 The whole effect of these four classes of disturbances as shown by the record 

 curves already mentioned will be, under usually good conditions, to displace 

 true deflections on an average about one-tenth millimeter on the plate. All 

 true deflections of a magnitude not greater than two-tenths millimeter are 

 lost among the false ones. 



III. Sources of error from imperfect mechanisms. 



1. Irregularities in the motion of the plate. 



Errors from this source negligible. 



2. Irregularities in the motion of the circle. 



Errors now occasionally as great as 3 seconds of arc; but steps being now 

 taken which promise to reduce these irregularities to a magnitude of five- 

 tenths second. 



3. Difficulty in correctly determining the ratio of plate and circle motions. 



A more accurate method of determination now projected will undoubtedly 

 give sufficient accuracy. 



IV. Sources of error in preparing cylindrics. 



1. Personal error in "blocking in." 



Experiments show that the mean displacements of lines from this cause vary 

 with three persons from 0.16 to 0.32 millimeter. 



2. Imperfect adjustments of camera. 



As a result of recent changes the error from this cause will not exceed one- 

 tenth millimeter in the length of a 20-ceutimeter plate, and is within certain 

 limits proportional to the length of the plate employed. 



V. Errors in comparator measurements. 



1. Personal error in the determination of a minimum upon the curves. 



Experiments show that the average deviations of the mean of two observations 

 by one observer from the mean result of six observations on the same curve 

 by three observers varies with different persons from 0.022 to 0.016 millimeter. 



2. Errors peculiar to the comparator. 



A new comparator is projected for use in the observatory in which it is 

 believed these errors will be negligible. 



'S. P. Langley, "Temperature of the Moon." Memoirs of the National Academy 

 of Science, Volume IV, 9, Appeudix 3. 



