The H2O+ Contents of Some Geochemical Standards Predicted by a Calibration Line

The combined water (H₂O⁺) in 30 geochemical standards having known and unknown water contents was determined by an elemental analyzer on 1 day during each of 6 weeks over a period of several months. The analysis of variance of the data measured in the form of a Youden square shows an extremely significant variation due to some unknown cause during the weeks in which the data were obtained. The higher water contents of the first two sets were obtained in the fall season when this area (Reston, Va.) has high and variable humidity, whereas the last four sets were measured during winter months. Humidity, however, was not included as a variable in the design and was not measured. The first two of three average blanks for H₂O⁺, measured after determining H₂O⁺ contents of about 11 percent for two samples, are noticeably higher than the third average blank. Similarly, the first of two average blanks for H₂O⁺, measured after determining the H₂O⁺ content (4.70 percent) of PCC-1, is higher than the second. Data for the last 4 weeks were used to calculate the line of regression and the extremely significant correlation coefficient. The line is used to predict suggested revisions of the “best” H₂O⁺ contents of the standards. Coefficients of variation calculated from the data for the last 4 weeks indicate that there is no severe sampling problem due to the small sample size (20 mg) taken for the determinations. The coefficients follow the general trend of large coefficients for samples containing the least amount of the constituent (BHVO-1: x?= 11.76% H₂O⁺; C.V. = 7.3%) and of small coefficients for samples containing the greatest amount (BX-N: x?= 11.76% H₂O⁺; C.V. = 0.4%).