NO. 3 MEASUREMENT OF OZONE — WULF AND ZIMMERMAN 7 



It should be possible to improve the value of rj in the course of a 

 program of ozone measurements, taking advantage especially of days 

 of high precipitable water. Each day, with the current value of rj, 

 should yield a new value of this quantity from the equation for Place 

 24 separately with the values of x, 8, and C for the day, and a running 

 average of t] may be kept. However, in attempting to do approximately 

 this in the course of the present work a further point arose which is 

 of some interest in the matter of method. 



With the adopted value of the ozone absorption coefficient for 

 Place 24 (section 6), essentially the excess of the left-hand side of the 

 equation over the right-hand side did not appear to approach zero 

 with decreasing water, but rather a finite positive value. The points, 

 indeed, fluctuated greatly and were not sufficient to approximate a 

 linear relation, but it seems clear that for the lowest water values the 

 apparent excess absorption is more than can be accounted for by 

 water alone. This means either that the adopted value for the ozone- 

 absorption coefficient at Place 24 is a little low compared with the 

 values at the other places, or that there is another, not so far con- 

 sidered, weak atmospheric absorption here. This matter is discussed 

 further in section 6. 



At the precipitable water values experienced in the present work, 

 the spectrum Places 22, 26, 28, and 30 do not appear to be influenced 

 by water absorption of intensity sufficient to be important in the 

 measurement of ozone. The transmission coefficients of Places 19 and 

 20 are determined from the envelope of the hologram trace in this 

 region (see p. 179 of footnote 7 reference). 



5. REDUCTION AND SOLUTION OF THE EQUATIONS 

 The set of seven equations under (2) above is of the form 



Cn = k„-X + ln'8 + 7itn't, (S) 



where the c„'s contain the observational data and certain known quan- 

 tities that do not change from day to day, and the coefficients kn, In, 

 and nin (all the irin's are unity) of the three unknowns x, 8, and ^, are 

 known quantities that also do not change from day to day. Given the 

 values of the transmission coefficients at the n places (in the present 

 work seven) in the visible spectrum, the n values of c, the left-hand 

 side of (2), can be readily computed since the other quantities con- 

 tained in c are at hand. (In computing C24 the value for the precipi- 

 table water at the time of the observations is also needed if this is high 

 enough to warrant correction for water absorption, as described in 

 section 4.) 



