1915] Michael, et at. : Hiidroyraphic Records of Scripps Institution 21 



tion, Brennt'ck made a nunihei- of experiments during the "Planet" 

 expedition (see Helland-Hansen, 1911-12, p. 7), which resulted in the 

 discovery that, when the inclination from the perpendicular was less 

 than 30°, the error was less than three per cent of the depth. By 

 manoenvering the "Agas.siz" we are visually able to maintain the cable 

 within an angle of 15°, which corresponds to an error of less than two 

 per cent. In most cases when the inclination exceeded 30° the collec- 

 tions were not kept, and in those few instances where they ^\■er(> kept 

 the error is noted in the tables. 



By comparing all the temperature-s and salinities relative to each 

 depth inve.stigated above two hundred meters we discovered a magni- 

 tude of variation exceeding that wliich would be expected from a 

 depth error of only two per cent. It would, therefore, seem that the 

 same angle of inclination produce.s an increasing relative error as the 

 depth decreases from two hundred meters. On the other hand, it is 

 certain that the temperature and salinity changes much more rapidly 

 in the shallower than in the deeper water. Furthermore, the effect 

 of drift and current in causing the inclination of the cable is less 

 during work in the upper levels, and the maneouvering of the boat 

 is much simpler. All things considered, the depth error probably does 

 not greatly exceed two per cent in any case. While this conclusion 

 applies to both soundings and intermediate depths, an additional 

 source of error, applicable to the latter only, arises from the occasional 

 failure of the water bottles to close before beginning to reel in the 

 cable. In such instances the temperature reading usually indicates 

 the failure immediately, in which event the haul is repeated. In 

 order to detect such occurrences which escaped notice on the boat, 

 all the temperatures and salinities relative to each depth investigated 

 were compared and, when both agreed with those from higher levels, 

 it was assumed that the apparatus failed to close at the proper depth. 

 As such data, on account of their misleading nature, are either ex- 

 cluded from tabulation or questioned, the error in depth relative to 

 the data published may be attributed entirely to that arising from 

 inclination of the cable. 



2. ERRORS DUE TO APPARATUS 



It will be readily admitted that it is impossible to construct any 

 piece of apparatus for the exploration of the sea that can be depended 

 upon to work perfectly every time. On the other hand, we doubt if 



