Subsampling Procedures

As mentioned previously, subsampling can occur both outside of the laboratory (to obtain a laboratory sample from the lot/batch) and within the laboratory (to obtain a test portion from the laboratory sample). In both cases, it is important 

Sampling of solid material often produces much more material than can be submitted to the laboratory for analysis. Reducing a large composite sample to a suitably sized laboratory sample often involves the following three steps  

A riffler is a mechanical device, consisting of a metal box containing a number of equally spaced slots, which is used for dividing a sample into two approximately equal portions. The material to be subdivided is poured into the top of the box and emerges through the slots on opposite sides in two approximately equal portions. As with coning and quartering, the procedure is repeated until the desired sample size is obtained. 

The analyst in the laboratory may also have to carry out one of the above procedures to obtain a suitable test portion from the laboratory sample. Prior to taking the test portion for analysis, the test sample may require additional treatment. 

Extraction of an analyte from a complex matrix, such as foods, is often dependent on the moisture and lipid content of the matrix. Hence, sample pretreatment 

---

The literature on sampling bulk materials is more specific and more quantitative than the general discussion presented above. Nevertheless, experiments on the sampling and subsampling procedures... 

The test presented in the previous section is useful when a smaller probability of false detection is needed than is provided by the distribution-free test. However, the test in the previous section is no panacea. Reduction of the skewness through proper choice of sampling and subsampling procedures is an alternative that may have much more potential for improving the study. 

In practice we have to limit the quantity of material, as most modern methods require only minimal amounts of samples. Usually we reduce the particle size by mixing. Frequently, as a measure for the introduction of the additional variation by the subsampling procedure the size-weight ratio is introduced. This is defined as... 

We will not go into which type of sampling is the most effective, but it must be emphasized that unless the correct sampling or subsampling procedure is used, the analytical data generated by the ICP-MS instrumentation may be seriously flawed because it may not represent the original bulk material. If the sample is a hquid, it is also important to collect the sample in clean containers (as shown later) that have... 

Single-tree plots of Jonathan, Delicious, and Winesap apples were sprayed on June 2 and 28 with 1 pint of Selocide and 1 pound of actual DDT per 100 gallons. Samples of one full box were selected from the north, south, east, and west sides and from the top of the tree to determine the effect of prevailing winds, amount of sunshine, etc., on the selenium residue. These samples were subsampled in the laboratory, stored, and analyzed by the same procedure as the samples in Table II. The results are presented in Table IV. 

Weliky et al. [154] described a procedure for the determination of both organic and inorganic carbon in a single sample of a marine deposit. Carbonate carbon is determined from the carbon dioxide evolved by treatment of the sample with phosphoric acid the residue is then treated with a concentrated solution of dichromate and sulfuric acid to release carbon dioxide from the organic matter. The carbon dioxide produced at the two stages of the analysis is estimated using a carb on analyser based on the thermal conductivity principle. In addition, total carbon content is determined on another subsample using the dry combustion furnace. This provides a check on the values determined by the phosphoric acid dichromate technique. 

An indication of the minimum size of a subsample can be obtained by using the concept of a sampling constant. For example, in the laboratory, the sampling constant can be used to estimate the minimum size of the test portion. However, the suitability of the chosen test portion size must be confirmed as part of method validation. The sampling constant Ks has units of mass. This is the mass of the test portion necessary to ensure a relative subsampling error of 1% (at the 68% confidence level) in a single determination. The value of /Ks is numerically equal to the coefficient of variation, CV (see Chapter 6, Section 6.1.3) for results obtained on 1 g subsamples in a procedure with insignificant analytical error. 

Subsampling can be done by hand, although several machines for subsampling are available. Another approach is to have an analytical method or methodologies that allow analysis of many different analytes in one sample. In this way, there is no question of variability between subsamples. This is not always possible, but minimizing steps in any procedure will decrease variability and uncertainty in the analytical results. 

McIntyre et al. [58, 59] conclude that the extraction of polychlorobiphenyls and organochlorine insecticides is most efficient at a total solids concentration of lg L-l, using the extraction procedure described above. The recovery of p,p -DDE from subsamples was always found to be the lowest of the four determinands considered (61.8%), while recoveries of Aroclor 1260, 7-HCH, and Dieldrin from the diluted sample averaged 96.3, 89.4 and 82.9% respectively at lg L-l total solids. 

Analytical Procedure. After thawing, each field sample was immediately divided into 4 quarters and subsamples of approximately lOOg of moist soil taken from each for analysis. A... 

The samples from the plot receiving the granulated formulation (Table I) revealed a different situation. The overall variability was much larger with a coefficient of variation of 395, with increased variability both between and within the five samples. Analysis of variance gave components of 140 between field samples, and 264 between laboratory subsamples. Since the analytical procedures were identical with those used in the EC plot samples, where reproducibility was good, these results clearly indicate a much greater irregularity of the distribution of the herbicide in... 

Chemical Analyses. Subsamples (<20 mg) of standing-crop-trap particulate matter were solubilized by acid digestion in sealed all-Teflon bombs (Bombco Inc.) in a procedure modified from Eggiman and Betzer (13). National Institute of Standards and Technology (NIST) standard reference materials River Sediment (SRM 1645), Urban Particulate Matter (SRM 1648), and mixed-element liquid spikes were used to check bomb performance. 

In this procedure, native substrates (triacylglycerols) are hydrolyzed to yield fatty acids. Subsamples are withdrawn from reactive mixtures at predetermined intervals, and reactivity is quenched by the addition of ethanol. The amount of fatty acids released during the reaction is determined by direct titration with NaOH to a thymolphthalein end point. 

Conceptually, assays for lipase activity using the colorimetric method (copper-soap procedure Basic Protocol 2) are similar to titrimetry in that liberated fatty acids are being measured however, the colorimetric method is more specific for fatty acids (Lowry and Tinsley, 1976). Quenched subsamples of emulsified acylglycerol/lipase reaction mixtures are combined with the biphasic mixture of cupric acetate/pyridine and benzene. Cupric salts of the fatty acids are formed (molar stoichiometry of fatty acid to Cu2+ of 4 2) and these soaps, which are blue in color, are partitioned into benzene to allow for quantification by measuring absorbance of the clear benzene phase at 715 nm. 

When identical lipase reaction subsamples are measured for fatty acid levels by both titrimetry (Basic Protocol 1) and colorimetry (Basic Protocol 2), estimates by the colorimetric procedure are only 60% of those obtained by titrimetry. However, the nature of this rela-... 

Aside from the time required to prepare reagents, the least amount of time is required per lipase assay by the spectrophotometric method, and the greatest amount of time is required per assay for the titrimetric method. Although all assays are described as requiring up to 30 min for the reaction mixture to be subsampled, time savings can be realized by subsampling more frequently over a shorter period of time, as long as one obtains valid initial rate data. Thus, for all assays, the time involved to run the lipase reaction can be normalized to be the same at -10 to 15 min. The difference in time requirements for the protocols becomes embedded in sample workup procedures. 

For the spectrophotometric method, there is no sample workup, allowing one to run -4 assays/hr. This can be increased to -16 to 100 or more samples/hr depending on equipment features and automation, such as multiple cuvette holders/changers and 96-well microplate readers. For the colorimetric procedure, sample workup requires -10 min/subsample, but several samples can be batch processed simulta-... 

---