THE PRECIPITATION OF CALCIUM CARBONATE IN THE SEA. 487 



It is thus obvious that none of these rnetals are suitable for the 

 work, and probably the only metal that could be used would be 

 Platinum, which would be prohibitive on account of the exp»ense. 



In order to overcome these difficulties, a water-bottle on a new 

 principle was desij^ed for me by Mr. Matthews. 



The container of the bottle consisted of a strong glass cylinder 

 holding about 250 c.c, : this was closed at each end by thick rubber 

 washers, through the centre of which a short piece of thin-walled 

 rubljer tubing passed, the tubing being sealed at the end within the 

 cylinder. The washers were fixed in metal plates sliding along the 

 guide bars of the skeleton frame in which the glass cylinder was fixed, 

 and by sending down two messengers along the sounding wire the 

 cylinder could first be opened at both ends and then closed at any 

 required depth. The whole apparatus was first sterilized by steaming 

 in a " Koch," and then the cylinder was completely filled with 95% 

 Alcohol : the washers were kept tight on the ends of the cylinder by 

 strong springs so that no leakage occurred- When the apparatus had 

 been lowered to the required depth, the first messenger was sent down, 

 this, by hitting a lever, opened the cylinder at both ends, and the 

 alcohol, being of lower specific gravity than sea-water, difi'uses out 

 almost in8tantaneou.sly, causing an upward flow of water through the 

 cylinder. On sending down the second messenger the cylinder, with 

 its sample of water, was tightly closed at each end by the rubber 

 washers. 



The washers, with their attached pieces of thin rubber tubing, had 

 sufficient capability of bulging inwards to allow for the contraction of 

 the Alcohol, due to the low temperature at any considerable depth, and 

 to its compressibility being greater than that of sea-water, and similarly 

 the expansion of the sample of water, as the apparatus was hauled up, 

 was compensated for by the partial collapsing of the thin-walled 

 rubber tubing. It is obvious that even liad a sKght amount of leakage 

 occurred, a leakage inwards during the descent of the apparatus would not 

 vitiate the results, as bacteria would promptly be killed in the 95% 

 Alcohol, and similarly on hauling up, the leakage, if any, would be 

 outwards, due to the expansion of the sample through the regularly 

 increasing temperature and decreasing pressure, so that the sample 

 would not be contaminated by any of the surface layers through which 

 it was hauled. There was, however, no reason to suppose that any 

 leakage occurred, and it appears that the expansibility of the rubber 

 washers and tubing was sufficient to allow for the small changes in bulk 

 of the fluids within the cylinder. After the first sterilization by 

 steaming, the action of the Alcohol was relied on for sterilization 



