TEMPERATURE AND SALINITY 



13 



boring stations. While the plotting of the observa- 

 tion points was mostly assigned to technical as- 

 sistants, the vertical curves were constructed with 

 the greatest possible care by scientific workers. 

 From these curves the temperatures and salinities 

 were taken with estimates to parts per hundred for 

 the standard depths, 200, 400, 600, 800, 1,000, 

 1,250, 1,500, 1,750, 2,000, 2,500, 3,000, 3,500, 4,000, 

 4,500, and 5,000 meters. For the older temperature- 

 series Professor Merz, himself, had completed the 

 work. These values together with the abbrevia- 

 tions of the names of the expeditions and of the 

 months of the observations, with the appropriate 

 isobaths, were plotted on surface-true charts on a 

 scale of 1:20 million; for the regions in which there 

 are more numerous observations such as the South 

 Antilles Sea, Newfoundland, and the west European 

 continental slope, special charts on a larger scale 

 were constructed. On the basis of the interpolated 

 values of temperature and salinity, the density 

 values were calculated, which because of the general 

 greater con.stancy at deep levels made possible 

 another test of the data. In the case of strongly 

 discrepant values it was possible in most instances to 

 decide whether the error lay in the measurement 

 of the temperature or in the determination of the 

 salinity, or whether in the construction of the 

 vertical curves insufficient attention was paid to the 

 correlation of Temperature-Salinity and whether a 

 .subsequent equalization of the curves for both 

 factors was necessary. It results from this graphical 

 investigation of the values at standard depths and 

 from the construction of horizontal charts, tempera- 

 ture, salinity, and density, that the curves may 

 not be constructed one independently of the other, 

 but that because of the essential relation between 

 the three factors every bulge in a salinity curve 

 necessarily requires a definite course of the tem- 

 perature curve, and that the density in depth should 

 show no irregularities. So far as it is attainable by 

 present state of the investigation, the charts of the 

 three factors must be drawn so that one is com- 

 parable with the other. Naturally it is not possible 

 to exclude all errors in working up so heterogeneous 

 material. Many bulges and peculiarities in the 

 isotherms, isohalines, and isopycnics apparently are 

 attributable to such sources of error. Among the 

 sources of error, above all other uncertainties that 

 manifest themselves, are those which result from 

 interpolation from observations with relatively 

 wide vertical observational intervals. These uncer- 

 tainties, especially in the temperature, exceed in 



most ca.ses all other sources of error in the measure- 

 ments; they are uncontrollable in amount if inter- 

 mediate maxima or minima occur between the 

 points at which measurements are made. 



A. TEMPERATURE 



In this presentation of facts it seems superfluous 

 to give a critical review of the methods of measuring 

 temperatures on the different ships, as has been 

 done for the measurement of bottom temperatures.* 

 Also for the present purpose the deep sea thermome- 

 ters used since 1873 may in general be regarded as 

 .sufficiently precise. With the old observations by 

 means of maximum-minimum thermometers, the 

 errors in measurement are mostly due to the subse- 

 quent displacement of the index. Systematic 

 deviations even in the depth of the intermediate 

 temperature maxima, where such deviations are to 

 be expected on account of the principle of measure- 

 ments on which the extreme thermometers are 

 based, are so in.significant that they play no role 

 on the horizontal charts. Rather is it necessary 

 to reckon with systematic errors in the measure- 

 ments with reversing thermometers of the old 

 construction in the years 1885 to 1905. Then in 

 most cases, because of the absence of an auxiliary 

 thermometer, the subsequent expansion of the 

 broken quicksilver mass was not eliminated from 

 the thermometric reading. The temperatures re- 

 ported by such research vessels as the Albatross, 

 Belgica, Princess Alice, and probably also in 

 part those of the Gauss and Pourquoi Pas, are 

 notably too high, particularly in the great depths 

 of the tropics and the subtropics. With the excep- 

 tion of the Meteor, Atlantis, and in part the 

 Deutchland, which controlled the depth of the 

 reversal by thermometric measurement, we must 

 consider in all serial measurements systematic 

 errors which result from the failure to take into 

 account the wire angle produced by the drifting of 

 the vessel. Because of the strong vertical gradients 

 there, errors due to this cause occur in numerous 

 serial measurements in the upper water layers of the 

 tropics. The strikingly high values which occur in 

 numerous series of the National, occasionally also 

 in those of the Berlin, Discovery, Margrethe, 

 and Valdivia can be attributed to too great wire 

 angle. Finally there remains to be considered that 

 the measurements were made in different months 

 and years since 1873, and that even the deeper layers 



' Compare, this volume, 1st Lieferung, pp. 12 et seq. 



