174 CHEMICAL COMPOSITIUN' OF RAIN. 



furnished to each of these centres as well as to the five laljora- 

 tories engaged in the chemical part of the work: — 



As a rule each month's supply of rainfall should be analysed at the 

 end of the month, a copy of all particulars in connection with such fall 

 being supplied, mentioning any possible source of contamination that may 

 have unavoidably occurred in the case of any sample either before, on, 

 or after collection. If the amount of rain, in exceptional cases, should 

 happen to be less than 200 c.c, the best course would be to keep it over 

 and add it to the next month's rain. 



After the method of collecting the rain, the method of 

 analysing it claimed consideration. In deciding on this, I con- 

 sulted the analytical chemists in charge of the variotis laboratories 

 outside Cape Town, anion g.st whom the work was to be distri- 

 buted, namely, at Bloemfontein, Durban, (jrahamstown, and 

 Pretoria. T had been furnished by Dr. Miller with a copy of the 

 methods in use at Rothamsted, but some of these seemed wholly 

 inapplicable under South African conditions, especially in certain 

 parts of the Cape Province. 



The following represents tJK- lines of procedure eventually 

 adopted. Determinations of nitrogen, both ammoniacal and 

 nitric, were regarded as of most importance, and, when there was 

 not sufficient water to enable other determinations to be carried 

 out, these were given the preference. When it was impossible to 

 devote 1,000 c.c. to these determinations, 500 c.c. was the quantity 

 substituted, or, if even that were not available, whatever quantity 

 was available was diluted with nitrogen-free water to 500 c.c. 

 before distilling. After performing these determinations, chlorine 

 was taken next in order. At Rothamsted the amount of water 

 prescribed for this determination was one litre, but here, again, a 

 smaller quantity had to suffice when necessity demanded it. I'or 

 the determination of sulphates, preliminary evaporation in a large 

 platinum dish was considered most satisfactory, but, in the 

 absence of such an appliance, a Jena glass retort or a large Jena 

 flask was decided on. 



As far as the actual analyses are concerned, it is i^lain that 

 uniformity of method was absolutely essential, seeing that the 

 work was being done at five distinct laboratories, and so the 

 following details of methods for chlorine, sidphur trioxide, am- 

 moniacal and nitric nitrogen, employed at Rothamsted, were 

 communicated to each of the^e laboratories, and adopted, except 

 as above indicated. 



Chlorine. — The method of estimation is that of titration with silver 

 nitrate, potassium dichromate being used as indicator. The amounts of 

 chlorine in rain show great variations, from one part per million to well 

 over 200 parts per million, but as a rule the proportion is too small for 

 accurate determination in the unconcentrated water. Prof. Warrington* 

 found that some rain waters give an orange tint with the reagents em- 

 ployed, producing the idea that chlorine is absent. This orange tint is 

 not discharged by the addition of a cliloride. The procedure adopted for 

 carrying out the determination is to add 5 c.c. of lime water, free from 



* Trans. Chem. Soc. i88q, vol. 55, p. .S45- 



