DETERMINATION OF ALTITUDES UY BAROMETER. 121 



TEST OF THE COMPARATIVE ACCURACY OF THE DIFFERENT METHODS OF COMPUTATION. 



On a former survey in Southern California, Lieutenant Williamson surveyed two passes, the 

 Tejon and the Canada de las Uvas, with a spirit level and a barometer, for the purpose of testing 

 the latter instrument. It was from these observations that Mr. Blodget deduced his empirical 

 table referred to above. In accordance with Lieutenant Williamson's request, I took the original 

 notes of the Canada de las Uvas survey, which he considered rather more accurate than the 

 other, and carefully tested by them these various methods of computing altitudes from barometric 

 observations. Before referring to the results obtained, a slight description of the pass will be 

 given. 



The Canada de las Uvas is a pass from the Tulare valley, in southern California, through the 

 Sierra Nevada to the basin east of the range. A better place for experimenting with the 

 barometer could not have been found, had this been tne sole object of the survey. The stationary 

 barometer was observed in a brush hut, at a depot camp situated near the head of the wide and 

 open Tulare valley, at an elevation of 1,447 feet above the level of mean tide at Suisun bay, 

 near Benicia, and distant 12.8 miles from the entrance of the pass. For about six miles this 

 pass is a narrow gorge, bordered by ridges several hundred feet in height. It then becomes an 

 open valley, from half to three-quarters of a mile in width. This character is preserved nearly 

 to the first summit, a distance of about 5.5 miles. The road then crosses several branches of 

 the Santa Clara, a river discharging into the Pacific, gains the summit of a second divide, 

 and descends to the basin, which is elevated about 1,500 feet above the head of the Tulare valley. 

 From the first summit to the basin, a distance of about 12 miles, the trail is bordered by low 

 rolling hills ; but the high ridges of the Sierra Nevada intervene between it and the Tulare 

 valley. 



By this description it will be seen that some of these test observations were taken in a narrow 

 gorge, others in a wide valley, and others in an open undulating country, separated by a high 

 range of mountains from the stationary barometer. The altitudes of the stations varied from 

 192 feet to 2,809 feet above Depot camp ; and their distances, from 12.8 miles to 36.4 miles from 

 the same place. A greater diversity in their positions could not have been desired. 



The observations in the pass were taken with one of Green's cistern barometers, similar to 

 those used on our survey. At the depot camp, another barometer of the same kind was used, 

 together with two syphon barometers, which, although greatly inferior instruments, furnished 

 a useful check upon errors of observation. 



In making the test computations, I prepared the table of abnormal corrections from the 

 observations at the depot camp. The table of horary corrections used was the one already 

 mentioned, deduced from Lieutenant Parke's observations on his recent survey in the vicinity. 

 The altitudes were first computed by the old method, with Lee's tables, using the carefully 

 interpolated simultaneous readings at the depot camp for the barometric and thermometric read- 

 ings at the lower station. To prevent the slight error arising from taking the ratio between 

 two equally erroneous barometric readings, the corrections for horary and abnormal error were 

 applied to the observations at both stations. The altitudes were then computed by the new 

 method, first using the observed, and then the mean daily air temperatures. As the results 

 were the heights of the stations above mean tide at Suisun bay, near Benicia, the altitude of the 

 depot camp above that level was subtracted from each in order to institute a comparison between 

 them and those determined by the spirit level and by the old method of computation ; both of 

 10 X 



