ACCURACY. The accuracy of analytical work varies within wide limits, according to the pur pose which an analysis is intended to serve. The most accurate analyses are those made to deter mine the proportions by weight, in which the various elements unite with each other. Thus, the proportion in which silver and chlorine unite forms one of the best determined constants of nature. In determining the proportion in which magnesium unites with chlorine, a series of de terminations has been obtained, agreeing so per fectly with each other that a loss or gain of only one-twentieth of a milligram of the magnesium chloride analyzed corresponds to the difference between the highest or lowest results and the average. No such accuracy is attainable in com mercial or technical work. Nor, if attainable, would it be of any value, since it is but seldom possible to obtain samples representing precisely the average composition of large quantities of material.
The aim of the commercial and technical ana lyst is usually- not to attain extreme accuracy, hut to obtain results which he knows to be cor rect within certain limits. Thus, if an analyst is required to find the percentage of copper in a sample representing a large cargo of ore, in order to fix its commercial value, he can deter mine the copper by the electrolytic method to within about one part in four hundred without undue expenditure of time or labor. If the ob
ject of the analysis is to .enable the superin tendent of the smelting furnace to make up charges of a suitable content of copper, a much quicker volumetric process is used ; the results are then less accurate than those of the electro lytic process, but still much more accurate than is necessary for the purposes of the smelter. When it becomes necessary to determine the amount of substances which occur in relatively very small quantities, it is impossible to avoid relatively large errors. For instance, in deter mining the amount of phosphorus in a specimen of steel• where the total amount is only about one part in a thousand, the analyst is not sur prised to find that, in spite of all care, differences of 2 per cent. occur between the results of deter minations made carefully and under exactly the same conditions.