DAVIS STRAIT AND LABRADOR SEA 17 



used was checked by using it in other titrations, all with no explana- 

 tion of the high salinities. The titrations were made within 24 hours 

 after collection of the samples. The sample bottles were of the 

 citrate of magnesia type and were well aged, having been used 

 throughout the season and in most cases having been used the pre- 

 ceding season. In filling the sample bottles from Nansen bottles the 

 sample bottle was half filled, shaken, emptied, and again half filled 

 and emptied before filling with the sample. New rubber washers had 

 been placed on all sample bottles just prior to the cruise. The main 

 valves, air valve, and petcocks of the Nansen bottles were repeatedly 

 inspected, and, as the temperatures are about normal, the Nansen 

 bottles have reversed at the proper level. As the vertical temperature 

 gradient in the laboratory is considerable and the samples are ordi- 

 narily stored on the deck, whereas the standard water is kept at a 

 level about 3 feet above the deck, the therinal expansion effect was 

 investigated by Mr. Alfred H. Woodcock of the Woods Hole Ocean- 

 ographic Institution by experiment, standardizing the silver nitrate 

 with Copenhagen water at room temperature and then measuring a 

 refrigerated sample by titrating it several times as it warmed up to 

 room temperature. As a result of this experiment it was concluded 

 that the error due to this source was probably less than 0.05%o in 

 salinity and certainly less than 0.10%o. 



A group of 42 samples, originally titrated immediately after col- 

 lection in July 1933 and which had been brought back were then 

 again titrated by Mr. Alfred H. Woodcock at Woods Hole in 

 October. These results averaged 0.018%o chlorine lower than the 

 first results, 32 samples freshening, 9 samples being saltier, and 1 

 sample being the same. However, the samples were allowed to stand 

 another 2 months and then were measured for a third time in De- 

 cember. The December titrations averaged 0.014%o chlorine lower 

 than the second measurements, 40 of the 42 samples being fresher 

 and two being slightly saltier (0.001%o and 0.002%o chlorine) 

 than found in October. We shall not discuss here the causes of this 

 continued freshening which averaged more than 0.011%o salinity 

 per month for 5 months; but, whatever the causes, the second and 

 third titrations, because of the relatively small salinity differences, 

 throw no direct doubt upon the first titrations, or at least not upon 

 the chlorine values found in the first titrations. 



The fact that during the fall of 1932 Wilson and Thompson (1933) 

 found a strong influx of salty water from the Atlantic in the deeper 

 layers on the Grand Banks indicates a flooding of the Gulf Stream 

 and suggests that the high salinities found in 1933 in the Labrador 

 Sea are not beyond the bounds of possibility. The axis of the highest 

 salinity water off the Greenland coast, according to the 1933 results, 

 coincides very well in location with the usual hi^h salinity axis and 

 grades off to small anomalies on the Labrador side, thus making it 

 impossible to deduct a constant amount from all the measurements 

 without making the salinities on the Labrador side abnormally low. 

 This lends credence to the 1933 observations ; but because the salini- 

 ties are so unusually high, and because there is not a corresponding 

 increase in temperature, the salinities have not been used except for 

 the construction of a dynamic topographic chart, and are not pre- 

 sented here in graphical form, but appear only in the tables. 



