ON THE TEMPERATURE OF THE FRESH-WATER LOCHS OF SCOTLAND. 429 



the air is not greatly different from that of the water. During frosty weather the 

 curves obtained were of a much more ragged nature than were the curves obtained in 

 more temperate weather. This is probably owing to convection currents. Towards the 

 end of September inverse stratification was noticed on one or two occasions, and this 

 gives an explanation of the raggedness of the curves representing the temperature at 

 the surface at this time. The surface water first cools down and then gets slightly 

 heated up by mixing with the warmer layers below it. 



Towards the end of May 1904, for three consecutive days the platinum thermometer 



was kept near the surface, and on each of these days the surface temperature varied 



more or less gradually. But in each case, in the evening, between 8 and 10 o'clock p.m., 



there was an abrupt change of temperature, followed by numerous rapid changes of 



temperature. On all three occasions the abrupt change was preceded by a rise in 



temperature of the upper layer water. The records for these three days are shown in 



fig. 14. The probable reason of these changes is, I think, to be sought from the fact 



that the surface temperature will have fallen considerably owing to the cold evenings, 



and so the water will have got into an unstable condition. Convection currents would 



then be set up, which would account for the rapid changes shown in the embroideries 



on the curves. The facts that the first sign of the changes which occur is a rise in 



temperature of the upper layers can only mean that the mean temperature of the water 



above 5 feet (the depth to which observations extended) is greater than the 



temperature of water at 5 feet, and that consequently when there is a tendency 



towards equalising the temperature of water to this depth, the water at 5 feet must 



rise in temperature. Later in the year, when the evenings are not so cold, and when 



accordingly the water does not cool off so rapidly, there are still indications of a like 



change, but at a much later hour, which is in agreement with the explanation which I 



have suggested. These convection currents, if they are such, do not reach to a greater 



depth than 10 feet, or at least there is no evidence of their doing so. The records of 



temperature at 1 feet at this time of year are rather scanty, but from the records which 



there are it would appear that it is only in the first 5 feet or so that these rapid 



changes at the surface are of any considerable amplitude. 



Towards the end of May the records of temperature at the surface began to show 

 rapid changes of great amplitude. So erratic did the curves obtained by means of the 

 Callendar recorder appear that they were at first attributed to instrumental errors. 

 But the changes were checked by means of mercury thermometers. It was not possible, 

 either, to attribute the changes to wind, for they occurred on the calmest days. Nor 

 can they have been due to river influences, for they were also observed at Dores, 

 where there is no river entering the loch to make the observations suspicious. On one 

 occasion, in two minutes the surface temperature was found to change as much as 6° F. 

 It is difficult to find any adequate explanation of this. On another occasion, when there 

 was a quantity of pollen from flowers on the shore suspended in the loch, it was 

 observed from the motion of the particles that different layers of water were moving in 



