reaching Yen-an Fu were hurried — owing to various circumstances — 

 and the survey got in consequence considerably out, i6' of Long, and 

 8^' of Lat. It was adjusted by the distance as measured by road-wheel 

 — incidentally showing to advantage this less ambitious apparatus — and 

 a new base was established as described above in (b), from which the 

 survey was begun afresh. 



At Lan-chou Fu, where I again joined the main body, the plane- 

 table results checked out to 12' of Long, and 6' of Lat. 



The map, which is based on the plane-table sheets, has been 

 adjusted to the correct positions of the various towns, as determined 

 by observations. 



Latitudes. 



The Latitudes in the Table (on Page 170) are always the mean 

 of at least two observations, and more often of from four to six. As the 

 greatest variation never exceeded 8", we think they can be taken safely 

 as correct. 



In taking the Latitudes for the ends of the base (b) the two 

 theodolites were used, both Grant and myself reading 30 faces on each 

 instrument. In one case the results checked to within i"5" and in the 

 other to o-8". 



Longitudes. 



By the co-operation of Major H. R. Davies, General Staff, who 

 kindly offered to take the Tientsin end, we were able to determine the 

 telegraphic Longitude of T'ai-yiian Fu. One set of six taps at 10 second 

 intervals, and another set of six at irregular intervals were sent and 

 received on three nights. Four persons received at T'ai-yiian Fu, and 

 two at Tientsin. The mean result was unexpectedly good, working out 

 to a probable error of 4 seconds of arc. 



In taking the Longitudes by occultation at Yen-an Fu, Hsi-an Fu 

 and Lan-chou Fu, never less than three observers took the time. The 

 greatest variation between the times of two observers was three-quarters 

 of a second. The immersions were always on the dark limb. These 

 results can be taken safely to a quarter of a mile. 



The Longitudes by Chronometer can, we think, be considered as 

 exceptionally good. All but that of F^n-chou Fu were between two points 

 fixed by occultation or telegraph. Two separate Longitudes of Yii-lin 

 Fu, one connected with the Longitude of T'ai-yuan Fu and the other 

 with the Longitude of Yen-an Fu checked to i '. 



132 



