a Van Dorn sampler were made at two depths, 5 m and 16 m. Two or three dissolved gas sampling bottles were filled with water from each of the casts with the Van Dorn sampler. The dissolved gas concentrations in each sampling bottle were determined three or four times. These data, therefore, indicate a measure of the variability encountered in analysis of one sample bottle and from taking separate samples at a particular depth. The range of values was greater for separate samples from one depth than for several measurements from one sample bottle. With one exception, the mean values for both nitrogen and methane concentrations from all sample bottles from a given depth were judged to be statistically equivalent at the 95% confidence level according to the t test (Huntsberger, 1961). The exception, the nitrogen concentrations for Van Dorn cast number one, was higher than those of casts numbers two and three. The variations in dissolved methane concentrations were slightly greater than those in the nitrogen determinations. In all cases, however, the precision of the analyses was within ±7% of the average and generally within ±5%.
ConclusionsThe data gathered by this study indicate that collection and storage of water samples in the dissolved gas sampling bottles is a method that maintains dissolved nitrogen and methane concentrations constant during storage. The overall method of collection, filling, storing, and analyzing is precise to within ±7% or less. The sampling bottles offer a convenient means of filling and transferring the water sample to the gas stripping chamber. When this method of sample collection and storage is used, the largest source of analytical error for this study was apparently the instrumentation.
Literature CitedHuntsberger, D. V., "Elements of Statistical Inference,"