Grab sampling usually consists in taking portions of ore at random from muck piles, chutes, or the tops of cars of ore. When done in this manner it is a haphazard method that can be relied upon only for roughly approximating the grade of the ore and often is employed merely for that purpose. The human factor or tendency to select for the sample too large a proportion of the best-looking pieces of ore, pieces of convenient size, or only the fines is more inherent in the application of this method than in other methods of sampling. Data on results of grab sampling at many mines indicate that usually they are too high.
The authors are of the opinion, however, that if grab sampling is done as systematically and with the same care commonly exercised in channel sampling, equally reliable results may be obtained, especially with ores of certain types. There are some ores in which the valuable minerals are contained almost entirely in rich patches or bunches in almost barren rock, like plums in a pudding, and that cannot be sampled accurately by channeling but that can be sampled with a fair degree of accuracy when broken by taking grabs from many points to form large samples.
As with other methods of sampling, the degree of refinement warranted by the purpose and cost of sampling should be considered.
The errors inherent in haphazard grab-sampling may be reduced by taking large samples—1 shovelful out of every 5 to 10 (or other proportion, depending on the nature of the ore and size of the individual pieces)—during loading into cars, or by using a scoop and taking one scoopful of ore from points equally spaced, by actual measurement, over the top of the pile (fig. 8). A convenient way to spot the points on a muck pile is to use a rope with knots or metal tabs at 1-foot intervals and stretch the rope across the pile on parallel and equally spaced lines. At each knot or tab, one scoopful should be taken for the sample. If the knot falls on a lump larger than the scoop will hold, a piece equal to a scoopful should be broken off under the knot with a hammer. This method may also be used for sampling loaded cars of ore, and when so employed samples should be more reliable when taken from the freshly loaded car before it has been trammed or hauled a long distance with resultant settling of the fines.
Either of these methods largely eliminates the human factor in grab-sampling by making it systematic, and although the samples of individual piles or lots of ore may be high or low, errors will become compensating as more and more lots are sampled in the same systematic manner.
Accuracy Of Grab Samples
Grab-sampling usually is employed where a high degree of accuracy is not expected, for control of breaking in stoping and drifting, as a check against channel or drill-hole samples, for determining the approximate grade of ore from individual stopes or chutes, or as a general check on the grade of mill heads.
At the Vipond mine, Timmins, Ontario, only a very small proportion of the gold is visible, by far the greater part being finely disseminated throughout the mass and favoring the pyrite, quartz, and schist in the order named. Diamond-drilling, channel-sampling of development faces and backs, stope faces and backs, and grab-sampling of the broken ore at each working place (stope or drift) all were employed. Grab samples consisted of a handful of fines from each car as it was loaded, and although this method is crude and day-to-day samples varied considerably, the results obtained were much more valuable in regulating the mill feed and indicating the grade of ore currently being obtained from various working places. The value of the ore milled during 1930, as calculated from the muck-(grab-) sample assays, was 4.8 percent below actual mill recovery plus tailing loss.
At the Wright-Hargreaves mine in the Kirkland Lake (Ontario) gold district, grab samples from muck piles and cars ran 30 to 40 percent high.
At the Douglas Island mines (Alaska), Bradley has summarized the experience with grab samples over a period of years as follows:
Jones has stated that in the Cripple Creek (Colo.) district grab samples taken from chutes (a handful from each car trammed) run 20 percent higher in value than the settlement value of the ore.
In the Mogollon district, New Mexico, gold-silver ore occurs in veins in a gangue of quartz, calcite, wall-rock fragments, and some fluorite. According to Kidder, a grab or “scale” sample was taken from each car as the ore was trammed to the mill and a composite sample made for each shift. At the end of the month the calculated average of scale samples was checked against the chute samples taken underground and against the average mill heads (calculated from production plus tails). On account of the slightly higher assay value of the fines in the ore, the average mine and scale samples usually were slightly higher than the mill-head sample. The annual average, however, seldom showed a variation of more than 10 percent, compared with the mill heads, and generally agreed within 3 or 4 percent.
Catron states that at the Colorada mine, Cananea, Mexico, results from grab samples taken from cars as they reach the station check the returns from the mill and smelter sample plants closely enough for purposes of control of production.
At the United Verde mine, samples of broken ore in the stopes or car or chute samples were necessary. It was determined that for such sampling, the proper-size pieces to choose were 3½ inches in all dimensions. Several cars of ore of approximately 1 ton each were chosen from various localities and sorted by hand into different sizes—minus 1 inch, plus 1 minus 3 inches, plus 3 minus 5 inches, plus 5 minus 7 inches, plus 7 minus 10 inches. Each size was weighted and assayed and the average assay determined for each car. It was found that the assay of the plus 3- minus 5-inch size was nearest that of the whole car, and the plus 1 minus 3-inch size second best, from which it was deduced that the proper size to choose must be slightly over 3 inches in diameter.
At the Eighty-Five mine, Valedon, N. Mex., mine samples (of copper ore) were generally 10 to 15 percent higher-grade than smelter samples.
At the El Potosi mine, Chihuahua, Mexico, experiments have proved conclusively that at this property the easiest and most efficacious method of sampling for general mining control is to take hand-grab samples from mine cars. The difficulties inherent in grab-sampling ores carrying appreciable lead values in the form of galena are recognized. Daily assays are made for silver, lead, and zinc, and the results are adjusted by predetermined factors for the three metals. The factors are obtained by comparing the weighted average of all assay returns during the month with the average mill heads, which are obtained from weightmeter and automatic samplers. Thus, comparisons may show mine-car samples to be 12 percent high in silver, 20 percent high in lead, and 5 percent low in zinc. For the following month the mill assayer applies these percentages as correction factors to the mine-car assays.