This section is from the book "A Treatise On Beverages or The Complete Practical Bottler", by Charles Herman Sulz. Also available from Amazon: A Treatise On Beverages.
Waters rich in magnesian salts possess laxative properties which should cause them to be rejected as a beverage, since their prolonged use may be injurious to health. If the proportion of the magnesium salts is higher than 2 1/2 to 3 grains to a quart, the liquid may be considered a mineral water. To free water from magnesia is a problem deserving serious consideration. After many trials the following process was adopted:
Treat the water in a tank with milk of lime, care being taken to agitate the whole from time to time. In this way magnesia, no matter how combined, will be precipitated in twenty-four hours.
Add to the water thus modified a certain quantity of finely pulverized witherite, or native carbonate of baryta, agitate frequently, and allow to settle down. All the lime present in the state of sulphate, that is the most of it, is precipitated after twenty-four hours. M. Reinsch, a distinguished German chemist, already employs with success witherite for purifying selenitic waters.
The proportions of magnesia and lime may, under various influences, vary for the same water, and as it is not practicable to estimate them chemically before each treatment, it may happen that an excess of lime or witherite will be added. In the first case, before beginning the second operation, it suffices to wait till the excess of lime has been turned into carbonate. This point is easily ascertained by means of test paper. The water must not be alkaline, or only very slightly so. In the second case, the excess of witherite, it must be borne in mind that carbonate of baryta, although insoluble in water, may become poisonous on being dissolved by the acids of the stomach. Hence it is absolutely necessary to filter the water. These operations may be performed in two barrels open at one end, one of the vessels being used for the manipulations, the other for the filtration. The last may be arranged in any way most familiar or convenient to each operator. For instance, over a layer of coarse gravel may be spread a layer of sand, then one of calcined charcoal, another of sand, and finally coarse gravel. At the lower part a faucet may be adjusted, and near it a vertical glass tube to allow the access of air. This general process is applicable to any kind of magnesian water, provided, however, the average proportion of lime and magnesia be ascertained, so as to know approximately the quantity of chemicals required. For selenitic waters, that is, those containing only sulphate of lime, the second operation alone is necessary. By boiling the water sulphate of lime or magnesia is also removed.
 
Continue to: