ATMOSPHEEE OF GREAT BRITAIN AND ON THE CONTINENT. 653 



on the 10th was due to the direction of the wind in the early morning. On the 

 15th the conditions were so very uncertain that observations for that day might have 

 been omitted with advantage. All the numbers that are too hierh in the table have a 

 note of interrogation placed after them. The observations, as entered in Tables IX., X., 

 XL, XII. and XIII., show very clearly the relation between the number of dust particles 

 and the transparency of the air. It is true the figures do not fit exactly into their places, 

 but it is not a little remarkable that they should fit as close as they do, considering 

 the difficulty of making estimates of haze, and also the probability that the particles will 

 not always be of the same size, which will influence their hazing power. And when we 

 further consider that most of the readings which do not fit into their places have to a 

 certain extent been accounted for, I think it will be admitted that the agreement is as 

 close as we are entitled to expect in an investigation of this kind. 



These tables show that the amount of haze depends on the number of dust particles 

 in the air between the observer and the object looked at, and the limits of visibility for 

 the different numbers show the different lengths of air required to produce the same 

 effect, namely, complete haze. From this it follows that the limit of visibility multiplied 

 by the number of particles ought to be a constant. This will be so, as it seems probable 

 that the same number of particles will produce the same amount of haze, whether the 

 particles be distributed through a long or a short length of air ; that is, the limit of 

 visibility will always be stopped by the same number of particles per unit of section 

 whatever the length of the column may be. In Tables IX., X., XL, XII. and XIII. 

 will be found, under column C, a series of numbers obtained by multiplying the mean 

 number of particles into the limit of visibility. It will be seen that the value of C in 

 Tables IX. and X. are fairly constant, considering the conditions and omitting the 

 exceptional readings obtained on the 4th July. In Table XL the values of C are 

 not quite so constant, but if allowance were made for the conditions under which 

 some of the observations were taken, as already explained, the agreement would be 

 much better. 



In calculating the mean value of C in Table X. the observations made on the 4th 

 July have been omitted, and in Table XII. the observations made on the 18th and 22nd 

 have been omitted for the reasons already given. 



If we compare the value of C obtained from the 1892 observations with C in the 

 1893 observations, we find that for wet-bulb depression of from 4° to 7° the agreement is 

 fair: the mean for C in 1893 is 105,923, while for 1892 it is 116,677, or 10 per cent, 

 higher. The agreement between the observations when the air was very dry in the two 

 years, as given in Tables XL and XIII., is not at all good. In 1893 C= 140,628, while 

 in 1892 0=174,832. The reason for this much higher number in 1892, as has been 

 already pointed out, is due to many of the dust observations in the 1892 table being too 

 high. It is on account of these 1892 observations not being so satisfactory as those 

 made in 1893 that they have been kept in separate tables, and we shall look on the 

 values of C as given by the 1893 observations as nearest the truth. 



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



