1863.] 519 



N.N.W. and E.N.E., and the barometer is half an inch lower with 

 S.W. winds than with N.E. winds. 



3. Temperature Temperatures between 30 and 40 will give 

 less ozone than when between 40 and 50, and the latter less than 

 when between 50 and 60. The same holds good when the ozone 

 box is artificially warmed. This does not extend to very high tem- 

 peratures, because the great dryness of hot weather is against the 

 action of ozone on the tests. 



4. Moisture. Increase of moisture up to a certain point is 

 favourable to the colouring of the tests, beyond which it operates 

 unfavourably ; for when the air is completely saturated with moisture, 

 the effect of ozone is at its minimum. 



5. Altitude. The higher the test is hung the darker will be the 

 colour obtained. The difference is as 4 to 6 between 4 feet and 

 35 feet above the ground. 



There are yet several other circumstances to be mentioned : 



1 . Hour of the Day. The difference between the ozone readings 

 at night and in the daytime are 



In June and July an excess at night of 0*1 

 In August and September 0'4 



In October and November 0*5 



In December and January 0*8 



In February and March 0*7 



In April and May 0*7 



the average excess of the summer months being only one-half of 



that which occurs in winter, 



2. Direction of the Wind. There is most ozone with the wind 

 between S. and S.W., and least when between N. and N.E. 



3. Protection of the test from light. It is absolutely requisite 

 that the test should be in a dark box ; and no box has been found 

 to answer so well as that constructed by myself, and known as 

 "Lowe's Ozone box," an account of which has been furnished to the 

 Royal Society* and to the British Association. This box, if freely 

 exposed, and made to veer with the wind, so as always to present the 

 opening to the direct current, is everything that could be desired. 



The foregoing observations will be sufficient to show that precau- 

 tions are requisite in these records, and that certain corrections are 

 * Proceedings, vol. x. p. 531. 



2r2 



