NOVEMBBK 17, 1922] 



SCIENCE 



557 



method, and no assumption need be made as 

 to tlie invariability of the coefficients of either 

 term. The means are thus well adapted to 

 show progressive or abrupt changes in latitude. 

 The annual rate is quite precisely the mean 

 rate for all the international zenith telescope 

 stations, +0.005" per year. A systematic cor- 

 rection to the proper motions, as originally 

 computed for their star list, would account for 

 this common rate. Its application gives the 

 last column of the UMah results, and its effect 

 is to diminish the average residual from 0.034" 

 to 0.027". 



LICK OBSEEVATOBY LATITUDE 



MERIDIAN CIRCLE 



Epoch No. Obs. cpo Corr. tpo 



1894.2 4 25.52" 25.52" 



5.2 4 53 53 



6.1 3 70 70 

 7.0 4 46 58 



8.0 4 72 57 



9.2 4 76 59 

 1901.1 4 83 74 



2.3 4 81 74 



3.2 4 74 67 



4.1 3 50 47 

 6.0 4 47 47 



7.3 3 .... 61 

 8.0 2 37 42 



12.6 4 64 64 

 13.3 2 50 50 



14.5 4 61 61 



16.7 4 .... 56 



17.6 5 50 55 

 21.6 3 .... 77 



Mean 25.60 25.59 



Av. residual ±0.12 ±0.08 



Preceding 1903 25.62 



Following 1903 56 



Preceding 1906 25.61 



Following 1906 _ 58 



Preceding 1911 25.59 



Following 1911 60 



Newcomb, 1901 to 1904 25.65 



Preceding 1901 25.58 



Following 1904 57 



URIAH, ZENITH TELESCOPE 



FOURTEEN MONTH MEANS 



Epoch (cpo) Eate 



1902.6 12.08" 12.12" 



3.6 12 16 



4.6 12 15 



5.6 11 14 



6.6 11 13 



7.6 11 13 



8.6 09 10 



9.6 15 16 



10.6 16 16 



11.6 16 16 



12.6 14 13 



13.6 13 12 



14.6 19 17 



15.6 24 22 



16.6 21 18 



19.6 16 12 



20.6 10 05 



Mean 25.14 12.14 



Av. residual ± 0.034 ±0.027 



Preceding 1906 14.107 14.142 



Following 1906 153 142 



Preceding 1911 14.117 14.139 



Following 1911 167 141 



Bate -f 0.005" per year. 



The probable error of a yearly zenith tele- 

 scope latitude is evidently about one third that 

 of the meridian circle results. There are ap- 

 proximately 3,000 observations per year in the 

 former, and 800 in the latter. Both classes of 

 c^bservations are subject to systematic errors 

 that produce larger errors in the mean results 

 than would be due to accidental errors of ob- 

 servation only. The purely accidental error of 

 a single meridian circle observation of a star is 

 closely ±0.2". With graduation error, and 

 the error of nadir reading included, the prob- 

 able error of a single latitude observation is 

 closely ± 0.3". The .single zenith telescope 

 observations have probable errors of about half 

 this size. The same list of stars was used 

 throughout at Ukiah, while the list varies for 

 nearly every year at Lick. 



There is no evidence at Ukiah of an abrupt 

 change in latitude at the epochs of the 1900 

 and 1911 earthquakes. An average difference 

 of 0.04" would be expected between any two 

 yearly results. Ukiah lies 26 miles east of the 

 1906 fault line, and Mount Hamilton is 22 

 miles east. These distances have generally been 

 concluded to be too large to show any indica- 

 tion of movement at the stations. The posi- 

 tions of the faults responsible for the 1903 

 and 1911 shocks do not appear to be on record. 



Without implying anything in the nature 

 of an apology for the qnality of modern as- 

 tronomical work, we must conclude that it is 

 hardly a criterion for such small changes as are 

 at issue in this case. The principle of natural 

 selection, useful as it may have been in the 

 domain of biology, must be used sparingly, if 

 at all, in astronomical results. This does not 

 preclude taking note of systematic errors, for 

 the existence of which there is evidence all too 

 ample for our purpose of as high precision as 

 we can reach. 



