102 DR HUGH ROBERT MILL ON THE 



fathoms, where the mean daily change is only half that at the surface, and then the rate 

 of change diminishes, until at 70 fathoms it is rather less than one-third as great as at 

 the surface, and appears to approach a limiting value. This curve appears to indicate 

 that no increase of the depth of Loch Fyne would suffice to absolutely prevent the 

 influence of surface change of temperature from affecting the deepest water. It furnishes 

 another proof of the necessity of caution in generalising from incomplete results ; for if 

 observations had only been carried to 50 fathoms, it would have appeared probable that 

 before 60 fathoms constant temperature would be reached. 



As in the Arran Basin, the air- temperature curve for Loch Fyne cut the surface 

 curve at the maximum, and the mean curve rather after the maximum. The theoretical 

 reason of this coincidence is obvious. The surface water continues slowly to rise in 

 temperature while the air is cooling down, as long as the actual temperature of the air 

 is above that of the surface, but as soon as the cooling air reaches the same temperature 

 as the water this rise ceases, and as the air becomes colder it chills the surface layers by 

 contact, and causes the first fall of surface temperature. On the other hand, the cooling 

 of the water appears to be checked at the minimum by the heating power of the sun, and 

 the air does not become warmer than the water until the latter has begun slowly to heat 

 up. As in the curves for the Arran Basin, the surface-water temperature in rising is, 

 after the air-curve crosses it, almost the exact mean between the air-temperature and that 

 of the mass of water. 



During the rise of temperature the air was warmer than the surface water for 165 

 days in 1886, 120 days in 1887, and 150 days in 1888, an average of 145 days; and 

 during the rest of the time the air was colder than the surface water for 216 days in 

 1886-87, and for 222 days in 1887-88, an average of 219 days. Speaking roughly, we 

 may say that the air is warmer than the surface water for four and a half months, from 

 the beginning of May to the middle of September, and colder for the seven and a half 

 months from the middle of September to the end of April. The number of days when 

 air was warmer than water for the two seasons 1886-88 were, for the Channel, Arran 

 Basin, and Loch Fyne respectively, 134, 136, and 143; while for air colder than water 

 they were 237, 228, and 219. This effect of isolation in lengthening the period in which 

 the air is warmer is evidently due to the lower maximum resulting from a slower rate of 

 rise of temperature, giving a longer interval before the falling air-temperature reaches 

 the same value and stops further rise of temperature. 



By interpolating probable values for the first three months of 1886, it is possible 

 to compare the annual mean temperatures of the two years ; and by interpolating probable 

 values for the last three months of 1888, the hypothetical annual means for three }^ears 

 may be compared. 



Hence each year the surface water was warmer on the average than the air, so that 

 it exercised, on the whole, a warming influence on the atmosphere, while the mass of 

 water, as a whole, was very little warmer than the air, and in 1888 appeared to be a 

 little colder. In 1886 the excesses of temperature of the surface water over the air in 



