8 FRIDTJOF NANSEN. M.-N. Kl. 



ways arrived at the same results. Two or three determinations were made 

 of most samples with different glass-chambers, and they nearly always 

 agreed very well. In all cases where there could be the least doubt as 

 to the values of salinity obtained, new determinations were made. 



A great number of water-samples have been re-examined with the 

 interferometer by my assistant, Mr. A. CYAN, this autumn and winter (1914 

 1915), two years after they were taken. The values of salinity found differed, 

 however, only very slightly from those found by the original examination. 



As a rule, or at least in the case of all important samples, the accuracy 

 of the salinity given in Table I may be expected to lie within + o.oi / o. 

 This method would easily allow of a much higher degree of accuracy; 

 the difficulty is, however, to acquire the sets of standard water for the 

 control, sufficiently accurately determined for the purpose. Exact deter- 

 minations of the specific gravity of the standard waters would naturally be 

 preferable to the determination of their chlorine. 



At Stat. 17, noo metres, Stat. 30, 400 and 440 metres, Stat. 36, 

 400, 500, and 580 metres, and at Stat. 57, 900 metres, two separate 

 water-samples were collected from the water-bottle for each of the men- 

 tioned depths. Examined with the interferometer these double sets of 

 samples gave values of salinity which agreed within a few thousandths of 

 one per mille for each depth. At Stat. 19 two water-samples were taken 

 in the same manner from 610 metres, but their salinities were found to be 

 34.916 /oo and 34.90 / o. The samples were first examined by Mr. J. 

 Helland, and a year later by Mr. A. Cyan, but with very nearly the same 

 results. It seemed at first dificult to understand the reason of this diffe- 

 rence. The explanation must be, however, that the samples were taken 

 with the Reversing Stop-cock Water-bottle, where the reversing thermo- 

 meter was placed inside the bottle, which therefore had to be opened at 

 the upper end (when reversed) and the thermometer had to be lifted be- 

 fore the temperature could be read off. During this operation a drop or 

 two of water may have fallen into the water-bottle from outside its upper 

 part when removed, or from the sounding-line; and as this water would 

 have been from the sea-surface, a drop or two of it might reduce the 

 salinity of the upper layers of the water inside the water-bottle appreciably 

 when added to it. Thus the last water-sample collected from the water- 

 bottle, and containing these uppermost layers, would get a lower salinity 

 than the first sample. This incident shows how very careful one ought to 

 be, when collecting the water-samples, in order to avoid that any drop of 

 foreign water be added to the sample. On the other hand it also bears 

 testimony to the accuracy of our determinations with the interferometer. 



