March 17, 1905.] 



SCIENCE. 



409 



nights at the Coast and Geodetic Survey 

 office. Sixteen observers took part in this 

 test. The observers were purposely se- 

 lected so as to include some with little or 

 no experience in any kind of observation, 

 some with long experience in astronomic 

 observations and in handling various in- 

 struments of precision, and some of various 

 grades between these two extremes. Two 

 observers worked at the same time, observ- 

 ing alternate stars, and thus obtaining a 

 determination of their relative personal 

 equation. One of the sixteen observers 

 was in the test continuously, became thor- 

 oughly accustomed to the instrument and 

 method of observation, and served as an 

 intermediary through which all the other 

 observers could be compared with each 

 other. 



The tests show that for a practised ob- 

 server with such a transit micrometer, the 

 total error for a star, including errors 

 which are constant for all the records as 

 well as the accidental errors of bisection, 

 is nearly the same for stars of all declina- 

 tions if expressed in angular measurement. 

 This is what should be expected if the 

 errors concerned are of the same nature as 

 if the object pointed upon were stationary 

 instead of moving. 



The accidental errors of bisection are 

 nearly the same expressed in angular meas- 

 ure for stars of all declinations up to 59°, 

 and are probably somewhat less for stars 

 of greater declination. This is an indi- 

 cation that the accidental errors of bisec- 

 tion are of the same nature as if the image 

 pointed upon were stationary, the indica- 

 tion being partly contradicted by the 

 smaller errors for stars of declination 

 greater than 58°. 



Good observations can be secured at once 

 with the transit micrometer without pre- 

 vious practise. Practise simply reduces the 

 accidental errors by about 25 per cent. 

 I feel that I may speak with assurance on 



this topic, for each of the sixteen observers 

 was forced to begin observing on the first 

 star that appeared in his field of view, 

 with no previous experience whatever. 

 This point is emphasized for the reason 

 that I had been led to expect that long 

 practise would be necessary before an ob- 

 server could be sent to the field with a 

 transit micrometer. The accidental error 

 of a single record with the transit microm- 

 eter is about the same as that of a single 

 record with a key. 



During the first half of the tests the 

 driving heads were geared to make one 

 turn in 2**. 4, when observing an equatorial 

 star. During the last half of the tests the 

 driving heads were geared to turn one half 

 as fast, namely, one turn in 4^.8. This ex- 

 treme change in speed produced surpris- 

 ingly little effect on the accuracy of the 

 result. With this instrument the speed of 

 4^.8 per turn, or possibly a slightly slower 

 speed, is believed to be most favorable to 

 accuracy. 



The tests show that the relative personal 

 equation between any two observers with 

 the transit micrometer is so small as to be 

 masked by the accidental errors of observa- 

 tion. This is equivalent to saying that it 

 is probably less in every case than ^.05, and 

 is, as a rule, much smaller than this. The 

 relative personal equation with a transit 

 micrometer is certainly not more than one 

 tenth as large, upon an average, as with a 

 key. This conclusion as to the relative 

 personal equation applies to inexperienced 

 as Avell as experienced observers. 



The literature of the transit micrometer 

 shoAvs abundant corroboration of these con- 

 clusions as to the relative personal equa- 

 tion. 



It is difficult to detect constant or sys- 

 tematic errors of any kind in transit mi- 

 crometer observations. All the errors 

 seem to belong to the accidental class. 



