high flows; with an abrupt transition 

 between the groups. Each set of data was 

 then adjusted with each individual alka- 

 linity vsilue being corrected for the ratio 

 of dilution between the flow at the time 

 of sampling and the mean ann\]al flow. This 

 is an inverse relationship. Logarithmic 

 plotting of frequency of occxirrence on 

 semi-log paper emd on log probability paper 

 showed the adjusted data to be geometri- 

 cally nonnal. Ccmparative adjusted alka- 

 linity values derived were as follows: 



Gecnetrlc mean; 



Standard deviation: 



AUtallnlty range containing 

 50^ of the obeervatlone ; 



U.S.G.S. 

 1910-11 



37 



3.11. 



17-80 



U.S.G.S. 

 19?3-5'' 



85 



1.22 



7'»-97 



Col . Rlv . Sur . 

 195't-56 



69 



1.1*1 



55-87 



On logarithmic probability paper, all 

 three plots overlapped in the highest value 

 range but were well gapped throughout the 

 remainder of the plot. The gaps between 

 the 191O-II values and the contemporary 

 values were large, indicating a significant 

 change in river alkalinity during the in- 

 tervening period that is not caused by 

 chance alone. The gap between the U.S.G.S. 

 1953-5^ plot and the 195^-56 Columbia River 

 Survey plot was small, the U.S.G.S. data 

 showing the highest values. These higher 

 vsLLues are caused principally by lower 

 river flows during the 1953-5^ sampling 

 period. These differences in alkalinity 

 are likewise sho^^n in the differences be- 

 tween the geometric means and the standard 

 deviations from the mean. The alkalinity 

 range containing 50 percent of the obser- 

 vations has narrowed greatly since I9IO, 

 indicating an increase in year-around alka- 

 linity values with the largest increase 

 occurring during the non-summer months (see 

 chapter on "YaJcima, River, Irrigation and 

 Pollutional Effects"). 



The standard error of the mean (S.E. 



= ^^'^ ■ 'v''^^ ' ) for the 1953-5^ U. S. Geolo- 



gical Survey's 36 samples = I.03. This 

 indicates, with other conditions being com- 

 parable, that the variation of 68 percent 

 of their yearly means, by chance alone, 

 will fall within the range of 82-88. Since 

 this is a reasonably narrow range, it 

 appears that tri -monthly analyses of compo- 

 site samples is a practicaliLe compromise. 

 To narrow this range down to 8U-80, 287 



yearly, or 2k monthly samples would be 

 required. This would almost require a 

 daily sample analysis which is impracti- 

 cable if a large number of sampling sta- 

 tions is involved. 



Conductivity and solids 



Solids or residue analyses are re- 

 ported as either total solids or separately 

 as suspended and dissolved solids (whose 

 sum equals total solids). In the vast 

 majority of samples tested, excepting for 

 Crab Creek, the turbidity was low and the 

 difference between total solids and dis- 

 solved solids was small. Total solids 

 only, were measured in this study because 

 of time limitations. 



Conductivity is closely related to 

 the dissolved ionized constituents in a 

 water (I6, I9) and can be used as a check 

 on the dissolved solids or total solids 

 (if turbidity is low) analysis. The test 

 for conductivity is rapid and precise, 

 whereas the test for solids is very slow 

 and subject to severe errors in sampling 

 or weighing. Over a period of time, ratios 

 of conductivity to solids can be estab- 

 lished for a given stream. This ratio can 

 be used to check the reliability of any 

 single solids determination. Figure 11 is 

 a plot of random conductivity and total 

 solids values obtained throughout the 

 Columbia River Basin. A straight line re- 

 lationship exists between the two. This 

 plot is slightly curved because the higher 

 values were for the Crab Creek area where 

 turbidity was high. If the Crab Creek 

 samples had been analyzed for dissolved, 

 and not total solids, the solids values 

 would have been lower, giving a straight 

 line plot. From Figure 11, it is deter- 

 mined that any single conductivity value 

 minus 50 can be multiplied by 0.7^ to give 

 the approximate value of the total (if 

 turbidity is low) or dissolved solids. 

 Using this relationship and comparing the 

 conductivity versus solids values in the 

 tabulations herein, it Is obvious which 

 solids values are probable in error. It 

 should be noted that the relationship is 

 of little value where the conductivity is 

 less than I50 micromhos. 



Hydrogen ion concentration (pH) 



These values were measured in the 

 field at the time of sampling with colori- 



