528 , . DONALD J. MATTHEWS. 



and the alcohol is displaced by the water. The second messenger, 

 which is hollowed out below so as to pass over the first, is allowed to 

 fall down the wire, and its broad base strikes the lever holding up the 

 shorter rod. This allows the upper washer plate to fall on to the top 

 of the cylinder, and the spiral springs keep the bottle tightly closed. 

 It is now hauled up and a sample removed by a sterilized rubber tube. 



The bottle is designed for use on a stranded steel wire, in the end of 

 which an eye is spliced. Through this eye passes a screw threaded 

 through two projections k shown on the upper side of the fixed plate ^. 

 The whole of the weight of the bottle and of any apparatus below it 

 falls on this plate, so brass sleeves j are fitted on to the frames below 

 the top plate to assist in taking the strain. A similar pair of cheeks 

 with a screw is fitted to the bottom plate to permit of other apparatus 

 being attached below. 



At is shown a rod which when the bottle opens drops till it is flush 

 with the lower plate c. A loop of wire slipped over this makes it 

 possible to release a reversing thermometer frame hanging below the 

 bottle, or a messenger to actuate other apparatus. 



The great defects to which water-bottles are liable are leakage and 

 closing at the wrong moment. During descent leakage inwards might 

 easily take place as the alcohol contracts on account of falling 

 temperature and rising pressure. To counteract this, the lower washer 

 has been made with a large dilatation ending blindly inwards but open 

 to the sea at the other end ; this would stretch slightly and compensate 

 for the change of volume to a certain extent. It seems, however, 

 to have been an unnecessary precaution. Leakage inward would be 

 so small that the alcohol would remain strong enough to kill any 

 bacteria which might enter, and could not affect the salinity of the 

 sample, as the water-bottle is thoroughly washed out when the first 

 messenger falls. Indeed, the escape of the alcohol is so rapid that 

 at a short distance below the surface the sudden precipitation of the 

 salts to which it gives rise has the appearance of an explosion. 

 Leakage during hauling up would be outwards, and a pumping 

 tendency by the rubber washers is not likely, as the water-bottle is 

 closed by springs and not by weights. 



That the water-bottle neither leaks nor closes at the wrong time is 

 shown * clearly by the sharp fall in the number of bacteria below a 

 certain depth, by the close agreement between the salinity at 400 

 fathoms at neighbouring stations, and by the agreement between the 

 ."^^alinities at the greatest depth at which it has been used and those 

 found by the Michael Sars at the same depth during her cruise in the 



* Drew, loc. cit. 



