40 DR HUGH ROBERT MILL ON THE 



November and December. Stating this generally, we may say that the water of the 

 Plateau is homothermic only just before the Spring minimum and just after the Autumn 

 maximum. During the months of heating (from April to August) the surface water is 

 usually more than two degrees warmer than the bottom layer, and may be as much as 

 8 degrees warmer. This is the most heterothermic condition assumed by the water. The 

 great warmth of the surface layers and its sharp contrast with that below may be 

 accounted for mainly from the seaward extension of the hot surface water of the Arran 

 Basin overlying the cooler water of the Channel. From the comparison of curves, it 

 seems that the deeper layers on the seaward side of the Plateau usually correspond in 

 temperature with the mass of water in the Channel. On the inner side, the Channel water 

 of the deep layers of the Plateau is distinctly chilled by contact with the colder deep 

 layers of the Arran Basin. 



Curves of negative slope occur from September to April, or during the season of cool- 

 ing, when the surface water is colder than that beneath. It is interesting to notice that 

 the negative slope on the Plateau was never so great as 2° in 25 fathoms, while the posi- 

 tive slope at times amounted to close on 8°. 



Fig. 7, Plate XXIIL, shows the typical curves of the four seasons — those of warming, 

 cooling, maximum, and minimum — on the Plateau. It will be noticed that the negative- 

 slope curve of cooling is much less pronounced than the positive-slope curve of heating. 

 It is natural to suppose that this may, to some extent, be due to the fact that all the 

 winter observations were made during the day when cooling by radiation is at a 

 minimum, while the summer observations, also made during the day, were at a time 

 when solar radiation was at a maximum. It will presently be shown, however, that 

 the positive form of the Summer curve is scarcely, if at all, modified during the night. 

 We may assume also that the short and infrequent sunshine of winter does little to 

 check loss of heat by surface radiation. The slighter slope of the negative curve is 

 more probably explained by the fact that the water on the Plateau is practically of 

 uniform salinity throughout, so that when the surface water is cooled its density is 

 increased, and it sinks, rapidly chilling the mass by convection, and so tending to produce 

 a homothermic state. The heated surface water, on the other hand, expands, and its 

 density diminishes by rise of temperature more than it increases by the rise in salinity 

 caused by evaporation. Hence the warm layer lingers on the surface, and the heat 

 passes downward slowly. 



It is noteworthy that on the Plateau smooth curves are practically never obtained. 

 At first sight, one would be inclined to attribute at least the minor irregularities to 

 observational error, but experience and repeated experiments have convinced me that 

 this is very rarely the case. A few of the more remarkable contorted Plateau curves are 

 reproduced in fig. 8, Plate XXIIL, in order to illustrate the manner in which they vary. 

 Nos. 14 (a), 2 (b), and 5 (c) show increasing degrees of complexity, due to the existence of 

 layers of water superimposed at different temperatures, an effect possibly brought about by 

 the complicated currents of the region, although, as will be shown immediately, instances of 



