six days, the pH dropped from 9 '05 to 

 Q.k, declining stesidily thereafter to 

 8.2. This decline csin be attributed to 

 the carbon dioxide (carbonic acid) 

 released on decomposition of the organic 

 matter in the sample. The carbonate 

 (C03~) alkalinity dropped 10 p. p.m. to 

 zero in the course of 26 days. These 

 ceirbonates were changed to the bicarbon- 

 ate (HCOo") by carbon dioxide in the 

 presence of water (003= + CO2 + H2O = 

 2HCO2"). TotEil hardness and total alka- 

 linity decreased 2 to 3 p. p.m. during 

 the 62 days. This slight decrease was 

 probably due to assimilation of these 

 constituents in the cell structure of 

 microorganisms and to precipitation. 

 (Samples were not shaken prior to each 

 determination) . 



Ammonia : This is largely produced by bio- 

 logical activity. Since ammonia deter- 

 minations are not practicable to run in 

 the field, the samples must be preserved 

 with sulfuric acid during their trans- 

 portation to the laboratory. In the 

 laboratory, they should be refrigerated 

 until the analysis can be made. 



Color : These deteiminations can be made in 

 the laboratory vinless iron or manganese 

 in any appreciable amounts are in the 

 sample in a soluble foim that will be 

 rendered insoluble on seration. Color 

 samples should be stored out of the 

 bleaching action of sunlight. 



Turbidity : Suspended matter in a sample 

 tends to settle and coalesce after a 

 period of several days. If then shaken 

 prior to a turbidity test, the particles 

 will not separate and give the same 

 turbidity readings as they would if mea- 

 sured within a day of sample collection. 



Total solids : Biological decomposition 

 will reduce the organic solids in a 

 sample if permitted to continue over a 

 period of several days. Total or organ- 

 ic solids should be determined within a 

 day or two of sample collection. 



Others: Determinations for sulfate, con- 

 ductivity and the various elements are 

 not appreciable altered through storage 

 of the sample prior to auialyses. 



RELIABILITY OF WATER QUALITY DATA 



The water quality of a stream is con- 

 tinuously changing. In a given stream, the 

 value of the constituent tested for will 

 vary with the rate of stream flow, with the 

 water use and with the air temperature or 

 season of the year. To obtain a reliable 

 documentation of the water quality, one has 

 the problem of determining how many and how 

 frequently water samples should be collected. 

 In their 12 established sampling stations 

 in the Columbia River Basin, the U. S. Geo* 

 logical Survey normally collects a water sam.- 

 ple each day. These samples for a ten-day 

 period are composited in ratio with each sam- 

 ple's conductivity. Thus, three constituent 

 values are determined during each month of 

 sampling. Even with these numerous san^jles, 

 there are abrupt changes at some stations in 

 the constituent values. The most accxirate 

 procedure would be the daily analysis of each 

 sample. This becomes a virtual impossihillty 

 when the nvmiber of samples and constituents 

 tested for are large. Collection of daily 

 saiqDles by a local resident of the area is a 

 good and an inexpensive way to get numerous 

 samples. It has the disadvantage of not per- 

 mitting a test for dissolved gases, ammonia, 

 phosphates, etc., and the samples have been 

 stored for a considerable period prior to 

 analysis (see section on storage of samples). 



On this contract, because of the large 

 number of sampling stations involved, be- 

 cause of the necessity of measuring dis- 

 solved oxygen, etc. at each station and be- 

 cause of a limited budget, it was not 

 possible to get frequent samples at each 

 station. Stations were sampled (composites 

 at each station of two or more Individual 

 samples) with a frequency of at least once 

 a month in the winter and up to ten times in 

 the summer months. To evaluate the reli- 

 ability of these samples with those collected 

 by the Geological Survey in 19IO-II and 

 1953-5^, a statistical analysis was made of 

 the alkalinity values obtained from the 

 lower YEikima River. (Available time would 

 not penult a more complete analysis.) 



Alkalinity values were avereiged for 

 each month of the year. For all three sets 

 of data, it was found that these alkalinity 

 values did not follow a normal arithmetic or 

 geometric frequency distribution. In a fre- 

 quency distribution. In a frequency plot, 

 the data divided themselves into two dlstiact 

 groups; those for low flows, and those for 



31 



