ATMOSPHERE OF GREAT BRITAIN AND ON THE CONTINENT. 661 



In this table, as in the others, the regular decrease in the limit of visibility with 

 the increase in dust is very evident, and the observations also agree very well with 

 each other. For instance, if we compare the results obtained on 24th September 1892, 

 the first observation in the table, with those taken on the 22nd September 1890 and 

 23rd September 1892, the two last observations, we find that the increase in the 

 thickness of the atmosphere is proportional to the increase in the dust, after allowance is 

 made for the greater humidity on the two last days. But when we come to calculate C 

 from these Callievar observations we get a result that does not agree at all with the 

 observations made at low level. From Table XVIII. it will be seen that C for Callievar 

 is only about half of what we obtained from the observations made at Kingairloch and 

 Alford. 



The Value of C at High Level. 



The much smaller number obtained for the value of C from the Callievar observations 

 casts a doubt on the value of these haze observations, or on the manner of working them 

 out. The first thing that suggests itself as the cause of the lower value of C at high level 

 is, that when we are observing at low level we may be testing locally impure air, which is 

 confined to a thin layer next the surface of the earth, while, when we are estimating the 

 haze, we are looking through only a little of this impure lower air and through much of the 

 purer upper air. Thus the number of particles observed at low level is too great, because 

 we have assumed that there are the same number through the whole air in which we 

 estimated the haze. On the other hand, at high level we count the number of particles 

 in the same air, or at least more nearly the same air, as we estimate the haze. The 

 value of C at high level ought, therefore, to be less than at low. The question then is, 

 what allowance ought to be made for this ? In fact, what is the difference between the 

 amount of dust at high and at low level ? 



Diagrams I., II. and III. show that there is generally less dust at high than at low 

 level. C, as obtained from the Kingairloch observations, is therefore too high, but it 

 would be difficult to say how much too high, as the difference varies from day to day, 

 and some days it is not too high, and occasionally it is too low. It would, however, 

 appear that the Kingairloch value is not too high to anything like the amount indicated 

 by the Callievar observations, because the air at Kingairloch is not liable to local 

 pollution, and the lower air is not much less pure than the upper, with the wind from 

 most directions. If, however, we were observing at, say, Baveno, where the air is locally 

 polluted, the result would be very different. We see from the observations made at 

 different levels at that station, given in Tables IV. and V., that there is a great difference 

 in the number of particles at high and low level. At 2000 feet there were on an average 

 less than half the number at low level. The value of C at Baveno would therefore 

 be much too great. We must not, however, place great weight on the Callievar 

 observations, as there are only three of them, the fourth being valueless on account of 

 the absence of humidity observations. 



VOL. XXXVII. PART III. (NO. 28). 5 G 



