497 



of clouds moving at a lower level than that of the true upper clouds. If this surmise is correct, it is 

 evident that had such clouds been grouped with the " lower " clouds, the change of the current towards 

 the south-west, already shown by the lower cloud observations, would have been even more marked, and 

 its further westerly development with height through the level of the Erebus smoke to that of the upper 

 clouds would have been still more apparent. But the figures as they stand in the table are sufficiently 

 striking: 64 per cent, of all the observations of upper clouds show a flow of air from between in nth 

 and south-west. At the lower level of the Erebus smoke 76 per cent, of the observations are comprised 

 between west and south, and, still lower, 47 per cent, of the lower clouds were found to move from 

 between south-west and south-east, whilst at the level of the sea 80 per cent, of the winds came from 

 between north-east and south-east. At the surface, east is the dominant wind direction, but ut the level of 

 the Erebus smoke, and of the upper clouds, the prevailing wind is from a westerly direction. 



VELOCITY OF THE WIND. 



The winds of that portion of the Antarctic region in which the "Discovery" wintered appear to possess 

 two distinctive characteristics exceptional gustiness and a great tendency to die away and give place 

 to calm. 



As regards the second of these characteristics, the observations of wind for the whole period of the 

 vessel's stay in Winter Quarters show an average of 23 per cent, of calms, the monthly percentages 

 varying from a minimum of 13 in August, 1892, to a maximum of 41 in July, 1903, both of them winter 

 months. But the distribution of calms over the different months is very irregular and shows little 

 indication of a seasonal variation. 



Away from the ship, on the sledge journeys, nalms were probably more frequent than at Discovery Bay, 

 but the irregular intervals at which the wind observations were made on these journeys, probably owing 

 to the exigencies of sledging, make it difficult to compare satisfactorily their relative percentage frequency. 

 The figures in Table V. show a frequency of 54 per cent, and 49 per cent, for calms upon the two western 

 journeys for which an analysis is given, and of 35 per cent, for the two journeys to the south. Each of 

 these journeys was of several weeks' duration, the shortest covering about fifty days, and the longest three 

 months. 



As has already been mentioned, the frequent occurrence of calms was a conspicuous feature of the wind 

 observations at Cape Adare and at the Winter Quarters of the " Gauss." 



The gustiness of the wind can be inferred from the fact that two numbers are almost invariably used in 

 the Journal to indicate the force as estimated by Beaufort's scale: e.g., " SW, 1-6; >S by W, 3-7." But 

 the inference is amply confirmed by numerous notes by the observers, who call attention to squalliness as 

 an especially characteristic feature of the gales and " blizzards " they experienced. 



MEAN MONTHLY VELOCITY. 



The mean hourly velocity of the wind for each month obtained from the readings of both the cup 

 anemometer and the pressure-tube anemometer is shown in Table XI. to be 10- 3 miles per hour by the 

 former and 10 '4 by the latter. 



It has already been said that the pressure-tube records of wind velocity were frequently interrupted, 

 and owing to that fact a satisfactory comparison cannot be made between the mean velocities for the 

 individual months obtained from the two anemometers, because the occurrence in any month of an 

 unusually windy period or of a long spell of very calm weather, during either of which the cup instrument 

 alone was at work, would obviously cause the mean velocity for the month deduced from the readings of 

 that instrument to be higher in the one case and lower in the other than the monthly mean obtained from 

 the smaller available number of pressure-tube observations ; and the rather large differences which the 

 table shows in the mean velocities by the two instruments for certain months is explained in this way. 



But as the length of the period covered by the records is increased the risk of error from this cause 

 becomes diminished, since the lacunse in the pressure-tube record extending over both calm and windy 

 periods tend to balance one another, and therefore it is probable that the mean hourly velocity fur the 



3 8 



