CLYDE SEA AREA. 63 



than that beneath, the effect of evaporation in this climate is not sufficient to increase the 

 density through increasing salinity so much as the increased temperature reduces it. 

 Hence the hottest surface layer tends to float and become still hotter, giving a very steep 

 positive slope to the curve. The mass of the water tends toward homothermicity by the 

 mixing effect of direct and indirect wind and tidal action, but the conduction of heat 

 downward from the warm, highly heterothermic surface layer gradually raises the 

 temperature, reducing the heterothermic layer, and giving a positive slope to the whole 

 curve. The greater the positive slope the more rapid is the flow of heat downward by 

 conduction, and the curve of density in situ becomes similar to a mirror image of that of 

 temperature, thus assuming a stable form and retarding the mixing processes which 

 involve vertical circulation. 



The mean difference of density determined at 60° F. between the surface and bottom 

 water at Skate Island was — 0"00052, the bottom being denser. This corresponds to the 

 ordinary state in hoinothermic conditions. For the maximum positive temperature 

 slope observed, viz., 9°'4, this difference is increased to —0*00219 ; while for the 

 greatest negative slope observed, viz., 2° -9, the difference is -0"00018. In the latter 

 condition it is apparent that the resistance to vertical movement on account of the 

 density of the layers, due to salinity and temperature, is less than one-tenth as great as in 

 the former, and a very slight increase of surface density would determine a downward 

 convection current. Inasmuch as the slope of a curve is estimated from the average 

 temperatures of layers of five fathoms thick, it is plain that the actual surface layer must 

 often be cold enough to cause an inversion of the density gradient and lead to downward 

 convection in a place like Skate Island, where the salinity gradient is so very slight. 

 These considerations fully explain the small negative slope, the parallelism, and the rapid 

 displacement negatively of the four curves for cooling, Nos. 5, 6, 7, and No. 1 of the 

 next year. 



The thermal conditions of themselves would bring about an approach to homo- 

 thermicity, and they are here reinforced by the other agents working in that direction — 

 the action of wind and tide. 



The time-changes of temperature are shown in fig. 1, Plate II., in the same way as 

 for Garroch Head, only, on account of the smaller number of observations, the horizontal 

 scale is reduced one-half. It closely resembles the Garroch Head diagram. The isotherm 

 of 50° was reached by the surface on July 8th, 1886, and had worked its way to the 

 bottom by October 28th. In cooling, the temperature of 50° was reached by the surface 

 on November 8th, and on the bottom by November 30th, the isotherm, which required 

 110 days to work down in rising, requiring only 21 days to work down in falling. Gauged 

 by this isotherm, the warm season at the bottom was one month, while on the surface it 

 was four. The isotherm of 54° reached the deepest point (7|- fathoms) on September 1 5th. 



In 1887 the surface was above 50° from May 28th to November 6th, a period of 5 

 months and 10 days, but the isotherm of 50° never reached the bottom at all, its utmost 

 penetration being to 62 fathoms on September 24th. The isotherm of 54° reached 8 



