Tin: MKUKl.NAL Sl'RlXtiS OK SOUTH AKRJCA. 529 



■thf dissolved hicarbonates of iron and calcium arc likcl\ u> l)t' 

 partly or entirely precipitated. 



Mineral springs, particularly new ones, occasionally undergo 

 considerable changes, even within comparatively short periods. 

 These changes may be ]>eriodical or permanent. A single analysis 

 cannot, therefore, safely be regarded as a criterion. It is 

 desirable to have a series of analyses extending over a term of 

 years until the composition has been proved to remain constant, 

 or the {)eriodicity has been established. An interesting example 

 is the Ehrlich Forest borehole outside Bloemfontein, of the 

 water of which several com])lete analyses exist. The results 

 show striking dififerences, although the analyses were carried out 

 within a few months of each other. The lithium chloride con- 

 tent in the last has decreased to the extent of approximately 

 1/35 of that in the first. In view of such variations future 

 analysts may find the inclusion of the date of each analysis in 

 the tables useful. 



As a rule, analytical results in this country are recorded in 

 terms of grains per gallon, and in the series of Tables No. I 

 I have kept the data in the original form. For purposes of com- 

 parison 1 have recalculated them in the usually adopted Conti- 

 nental standard, grams per 100,000 (series of Tables No. II). 

 There is a growing tendency among chemists, although not yet 

 universally adopted, to discard the conventional form of express- 

 ing results by combining anions and cations more or less arbi- 

 trarily, and to express restdts instead in terms of ions. Such 

 a system is perfectly leji'itimate, as the amounts of saline con- 

 stituents in waters are usually so small that they may be regarded 

 as completely ionised. This method (which has been ado])ted 

 in calculating the values in series of Tables No. II) does away 

 with the factor of uncertainty at present attaching to all water 

 analyses, inasmuch as the combination of ions is largely a matter 

 of choice. 



The value for total solids obtained by actual estimation is 

 usually different from that obtained by addition. It is for that 

 reason that the calculated and determined values are given in 

 different columns. 



Most analyses contain the values of iron and aluminium as 

 Fe^Og -|- AI2O3. In the calculation for the ionic Tables (Series 

 Nc. II) these values have been regarded as being pure Al. O3 

 on the one hand, and Fe-pO. only on the other. The values for 

 aluminium and iron calculated on this assumption are always 

 recorded in brackets. 



Silicic acid is probably present in the uncombined state in 

 most waters. Being an extremely weak acid, it must be re- 

 garded as practically undis.sociated. Nothing is known about 

 its state of hydration (ortho- or meta-silicic acid), and rather 

 than make any arbitrary assumption, I have preferred to leave 

 the values in the form usuallv given, namely, SiOo. 



From the value of combined carbon dioxide it is imixissible 

 to discriminate between carbonate and hvdrocarbonate ion. The 



