1916] Micliavl. ct al.: Hijdrographic Hicords of Scripps I iistifnlion 215 



a. Reasons for rejecting temperatures taken prior to June, 190S. — 

 During the preliminary explorations made between August, lf)()l, aiul 

 June, 1908, about seventy-five surface temperatures were taken with 

 ordinary chemical thermometers. Through oversight, their readings 

 were recorded without making the necessary calibration correction, 

 and. unfortunately, the thermometers were either lost or broken before 

 August. 1908, when it was discovered that the recorded temperatures 

 were subject to calibration errors. At this time, calibration of the 

 thermometer that had been used during the two preceding mouths 

 revealed an error ranging fi'om 0?9 to 1?5 C. thus indicating that all 

 seventy-five temperature observations were subject to calibration 

 errors approximating two degrees Ontigrade. 



Even so, the records might have been of some biological value were 

 it not for serious discrepancies between the positions of the tempera- 

 ture observations and those of the corresponding biological collections. 

 The magnitudes of these discrepancies, for the most part, are unknown, 

 but the records indicate that they must have frequently exceeded five 

 miles. Since differences of two degrees have often been observed 

 within a distance of two miles (MeEwen, 1916), it follows that, for 

 biological investTgations. these preliminary temperature records are 

 subject to an erratic error whose maximum approximates the observed 

 annual range of seven degrees. Needless to say. such misleading 

 temperatures are worse than none, and accordingly were not published 

 in our former (1915) report. 



b. Methods of detecting and rejecting doubtful subsurface observa- 

 tions. — In addition to mistakes which are not entirely avoidable, there 

 are two sources of troublesome error affecting the measurement of 

 temperature, and two affecting the measurement of salinity. Even the 

 best reversing thermometers occasionally register incorrectly because 

 the mercury column does not always break at the constriction. Like- 

 wise, the best containers may occasionally permit sufficient evapora- 

 tion to increase the salinity materially if the water samples are not 

 tested soon after being collected. While such errors are usually large 

 enough to be revealed as soon as the measurements are made, they 

 may be so small as to require for their detection special examination 

 of all temperatures and salinities of the corresponding series. Finally, 

 the collecting apparatus is designed to operate on a vertical cable, 

 and, in case of undue inclination of the latter, the thermometers may 

 either not reverse at all or may register at some higher level ; similarly, 

 the water bottle mav either fail to close entirelv or mav close at a 



