62 DR HUGH ROBERT MILL ON THE 



divergence in the upper 30 fathoms, showing a distinct paraboloid form, except that the 

 upper end shows the 5-fathom layer of warmest water to be comparatively uniform in 

 temperature. No. 3 shows homothermic heating throughout the whole depth below 20 

 or 30 fathoms, the straight vertical curve being pushed forward nearly 3° in 1886 in 



63 days, 2° in 1S87 in only 37 days, but only about 0°"5 in 1888 in 36 days. The upper 

 part of the curve is typically paraboloid, diverging positively from parallelism with the 

 preceding curve toward the upper end. In 1888, however, the upper end presented an 

 inverted paraboloid form ; which, together with the much slower rate of deep heating, 

 showed some exceptional retardation in the progress of annual change. This may be 

 related with the air-temperature (see curves, fig. 3, Plate XXII.), which was practically the 

 same in 1887 and 1888 for April and May, but in June was nearly 5° lower in 1888 than 

 in the preceding years. This lower air-temperature, or the causes which produced it, would 

 account for the inversion of the upper part of the 1888 curve, but not for the slow rate 

 of homothermic heating, which must be otherwise explained. In large part, the closeness 

 of the three curves is due to the exact minimum not being represented. Curve 1 shows 

 cooling still in progress, and curve 2 heating from the surface, while the lower part is 

 colder than in 1. Curve 3 in 1887 shows a distinct approximation to the sickle shape. 

 The intermediate minimum indicates that the rise of temperature had been less rapid at 

 50 fathoms than anywhere else. If, as it seems reasonable to assume, homothermic 

 change of temperature is effected by the complete vertical mixing of the water through- 

 out the homothermic depth, while the positive or negative heterothermic condition of the 

 upper layers is produced by positive or negative surface heat-exchange taking place too 

 rapidly to be equalised throughout the mass by the movements in progress, it would 

 appear that there was a freer circulation in the lower than in the upper half of the mass 

 of water. This might result from a peculiar arrangement of the salinity of the water, or 

 from a peculiarity in wind disturbance, but, unfortunately, no intermediate samples of 

 water were collected on the occasion. In curve 4 we see the typical paraboloid form of a 

 rapid rise of temperature through surface-heating. The homothermic condition has been 

 entirely overcome, and from bottom to surface the water grows warmer more and more 

 rapidly. Curve 5 (only properly shown for 1886) marks the passage of the annual 

 maximum. The curve is negative, for rapid surface-cooling has set in, and it approxi- 

 mates to the sickle shape possibly because heat is still being passed down by conduction 

 in sufficient amount to partially overcome the tendency to homothermicity. The slope 

 is, however, very slight. The curve in question shows a check to surface-cooling at the 

 time of observation by the very slight slope of the upper 10 fathoms. Compared with 

 curve 4, it shows most rapid cooling to have taken place at the surface, and most rapid 

 heating at the bottom, while at 17 fathoms the temperature w T as the same in both cases. 



The process of transition between the forms 4 and 5 is of very great interest, and may 

 be indicated thus. Until the equinox, there is a large positive gain in the surface heat 

 exchanges of the water by solar radiation, and until about the middle of August by 

 contact with warm air. The surface density being much lower, through its less salinity, 



