8 FRIDTJOF NANSEN. M.-N. KI. 
different cases, and that consequently about the same quantity of ice may 
have been formed on the bottle. It is also a striking coincidence that the 
salinities obtained are especially high on days with very low air-temperature 
(e. g. May 3, 6, 26—28). The glass-bottle may probably have been very 
cold, when it was sent down, and the cold glass with frozen hempen net- 
work outside, may have intensified iceformation inside the bottle. In some 
cases the sample taken from one depth (e. g. 5 metres) gives a very high 
salinity, whilst the samples taken from other depths give more probable 
salinities. The latter may have been taken after the former, and in that 
case the bottle by being filled with water once before, may then have 
become heated to the freezing point of the sea-water; with a smaller forma- 
tion of ice subsequently as the result. 
This formation of ice has, however, in several cases, obviously had 
the opposite effect, and produced samples which have given much too low 
salinities. The explanation is evidently that there have been numerous 
small ice-needles floating in the sea-water, which have got into the water- 
samples. We thus see that water-samples taken on such cold days, with 
the sea-water at about its freezing-point, are not trustworthy, unless special 
precautions be taken to avoid errors caused by the formation of ice on 
the water-bottles. Even samples taken with the bucket on such days, may 
give erroneous results from the same reason; but, as the water-bucket 
would hold a much greater quantity of water, and the hauling up would 
take less time, the effect of the formation of ice will be much less in 
this case. 
The Water-Samples. 
The water-samples were taken very carefully especially from the 
deeper strata. The water-bottle was wiped outside and around the tap in 
order to prevent possible drops of surface-water from getting in with the 
water-sample when run out into the glass-bottle. The glass-bottles, used 
for surface-samples, would hold about 100 cubic-centimetres; whilst all 
samples from the Stations were taken in bottles holding 150 cubic-centi- 
metres. They were closed by cork-stoppers which had been carefully 
selected. The bottles had been washed out in hot water for several days 
at the laboratory, before the expedition started. When the samples had 
been taken, the cork-stoppers were driven down as hard as possible into 
the necks of the bottles, which with the corkstoppers were dipped into 
melted paraffine-wax, and tied over with fairly airtight paper. To judge 
from the determinations of the chlorine, especially of the samples from the 
Stations, it seems as if the cork-stoppers have on the whole held very 
