use in one particular liquid, and at a pretty constant temperature, fairly exact results are obtainable (within ·0005). 2. Instruments which measure loss of weight in a liquid.—Of these the Westphal balance is the most important and widely used, and is extremely quick and convenient, where great accuracy is not required, though open to grave errors. Its faults are the errors due to cohesion (very slight); to the difference between true weight and apparent weight in air; to the condensed film of air on the plummet, which is, however, removed on immersion in most liquids; to the unequal expansion of plummet and liquid of which the density is to be taken at any other temperature than that for which the instrument is set, and of the arms of the balance; to the wire or thread used to suspend the plummet (usually reduced to a minimum by the use of a very fine thread); and to the imperfections in the balance (often large in those commonly in use). Many of these errors may be neglected in ordinary work, for instance, those due to cohesion, unless the fluid is very viscous; to difference between true weight and apparent weight in air, if the substance has a density near 1; to the condensed film of air; to the use of a wire or thread; and to the expansion of the arms. It is in the construction that the Westphal balance chiefly errs, one error having been pointed out by Allen (ANALYST, 1889, p. 11) in the length of the intervals between the divisions. The beam of the Westphal balance is usually made 10 centimetres in length, so that to insure absolute accuracy the intervals should be 1 centimetre apart, within of a millimetre, an amount of accuracy which it is not easy to attain in practice, and which may easily be diminished by wear. Another error is introduced in the hook riding on a knife edge, from which the plummet hangs; if the position of this is changed, as it must necessarily often be, a difference in the reading may be observed, amounting sometimes to ⚫001. Another error may occur from one of the riders not sitting absolutely exactly in its place, but inclined to one end of the beam or the other; this error may amount to as much as 0005. The error introduced by the expansion of the plummet is of course obvious, and should be allowed for by noticing the difference between the actual density of water at various temperatures and the indicated density. Unless this is done, determinations made at higher temperatures are only relative, and cannot even be compared with those obtained by another instrument, unless the coefficient of expansion of the material of which the two plummets are made is the same. An inconvenience is felt if the density of a liquid is to be taken at a temperature differing from the temperature at which it is a comparatively large bulk of liquid having to be warmed or cooled as the case may be. I should fix the limit of accuracy of the Westphal balance at not much less than 0005, although closer duplicate readings are often obtained. 1 100 The torsion balance is open to all the faults of the Westphal balance, except those due to construction, and besides has the tendency to acquire a slight permanent set, and is not to be recommended except for rough density determinations. 3. Instruments in which the mass of known volume is taken, or pyknometers.— These are practically two in number, the specific gravity bottle and the Sprengel tube and its modifications, neither of which presents any real advantage over the other. It is perhaps rather easier to adjust the temperature of the liquid in the Sprengel tube, but, on the other hand, the Sprengel tube is more liable to lose weight than the bottle, when in constant use. The errors, all of which can be readily allowed for, are the following those introduced by the condensed film of air, both inside and outside; by the difference between true weight and apparent weight in air; and by the expansion of the glass or other material of which the instrument is made. The most difficult to allow for is the condensed film of air, which I consequently try to eliminate as far as possible by the following plan: dry the bottle or tube by heating it and passing a current of dry air through it (of course removing the thermometer if the temperature is higher than the highest it will bear), cool, by pouring a current of cold water on the outside, wipe dry and weigh at once; duplicate weighings very rarely differ by as much as '0005 grm. I then fill with distilled water freshly boiled in a platinum vessel, adjust the temperature accurately to 4° C., wipe dry and weigh at once. The increase of weight in grams., corrected for the volume of displaced air, will give the number of cubic centimetres the bottle or tube will hold. The difference between duplicate weighings should not exceed 0005 grm.; the difference between the capacity at 4° and the weight (corrected) of water in grms. held by the bottle or tube at any other temperature, divided by the density of water at that temperature, will give the increase of capacity due to expansion of the glass or other material of which the bottle or tube is made. table of capacities at various temperatures, and a formula for the expansion, may easily be made of it for each instrument, TABLE OF DENSITIES OF WATER. (Compiled from the results of Pierre, Despretz, and Kopp.) A Should the capacity of the instrument be required at 100° C., some rather special precautions must be taken; the tube should be filled hot, and while the water is absolutely boiling, and the liquid should be adjusted to the proper level, while under slight pressure. I accomplish this as follows (with a Sprengel tube): To the end of the tube from which the liquid is drawn, a tube, about 1-2 c.m. in diameter, having its other end narrowed, and to which a side tube is attached, is fixed by means of a cork; a rod, on the end of which a bundle of filter paper or other absorbent material is fixed, is passed through the narrowed end in such a way that the joint is air-tight (i.e. either through a cork or india-rubber); the side tube, and the other end of the Sprengel tube are attached to a bottle, by means of a T-piece, in which a pressure of about ten inches of mercury is kept up by suitable means; the Sprengel tube is, after being filled, immersed in steam, and when the water ceases to expand the rod supporting the filterpaper, is brought to the end of the tube and by its means the water is accurately adjusted to the mark; the tube is then taken out and weighed in the usual manner. 66 The determination of densities at higher temperatures than 15.5° is usually only required for body that are solid at that temperature, such as butter, and other fats; Muter proposed to do this at 100° F. (37-8° C.) and graduated his instruments with water at the same temperature, and gave to these determinations the name of actual density," an unfortunate name, as they were really the actual densities divided by the density of water at that temperature. These, however, were good determinations, as there could be no mistake as to what they really were. Three years ago Estcourt proposed to take density of fats at 100° C. by the Westphal balance, a plan which has become rather general among English analysts, but is a distinct falling off from Muter's method, because the expansion, which is not usually allowed for, is enormously increased at that temperature; and moreover the temperature is exceedingly difficult to obtain, as the boiling point of water depends on the pressure of the air. I do not consider that the results obtained by two analysts taking their densities thus are comparable within 002; the only advantage which this method has over the other, is that beeswax and carnauba-wax are liquid at 100° C. while solid at 100° F. In the regulations for analysis for the State of Colorado, it is laid down the densities of fats shall be taken at 40° C., which seems to me the most convenient temperature, now that the Centigrade scale is in general use, and I think that we may expect, if the precautions I have drawn attention to are taken, that two analysts should not differ by more than 0002 or 0003. There are two minor precautions I should wish to draw the attention of analysts to, first, to allow a sufficient time (when using the bottle or tube), for the liquid to contract to its full amount; alcohol contracts very quickly, glycerol very much more slowly, while as an extreme case Perkin has recorded a mixture of aldehyde and water CHO+HO which took several hours (J. Chem. Soc. 1887, p. 817); the other precaution is to use an accurate thermometer, the great majority of low-priced thermometers being sensibly wrong. In conclusion, I would recommend that all densities should be the weight (mass) of the substance in the cubic centimetre, that the density of liquids should be taken at 15, and of fats at 40°. A symbol, such as D meaning density at 15° and Dio at 40 might be advantageously adopted. Some results actually obtained are appended. 159 I. Experiments showing the unequal expansion of specific gravity bottles and tubes: ON THE COMPOSITION OF MILK AND MILK PRODUCTS. BY DR. P. VIETH. (Read at Meeting February, 1889.) WHAT I have to bring before you to-night is my annual report on the work done during the year 1888 in the laboratory, which is under my charge. Particulars with regard to the purpose for, and the way in which the work is carried out may be found in my former papers of a similar nature (see the ANALYST, VII., p. 53; VIII., p. 33; IX., p. 56; X., p. 67; XI., p. 66; XII., p. 39; and XIII., p. 46), and need not be repeated on the present occasion. The total number of samples submitted to analysis during the year 1888 is 20,248, comprising In addition to these analyses, the specific gravity of about 75,000 samples of whole and skim milk was determined. Of the milk samples 12,682 were taken from the railway churns on their arrival in the dairy, and 2,545 by the inspectors employed by the business, from the men while the latter were serving the customers. The following table contains the monthly averages of the results referring to these samples : I may safely leave it to you to draw your own conclusion from these figures. A further number of 1,806 milk samples was analysed in connection with a very extended investigation into the composition of milk yielded by individual cows kept at the Aylesbury Dairy Company's estate, near Horsham. The results of this interesting investigation, which was commenced in the year 1887, will form very valuable material, which I hope to be able to lay before you on some future occasion. As in the case of milk, so in the case of cream, supplied to customers, two series of samples were examined, viz., 412 samples taken before the cream was sent out, and 304 samples taken by the inspectors from the men when working their rounds. The results were as follows:- The agreement between the two series of samples is satisfactory, considering the difficulty of drawing a fair average sample of cream of such richness, and further considering that cream sent out with the morning delivery was sampled before sent out, but not on delivery, being handed to the men, and by the latter to the customers, in sealed cans. The composition of 55 samples of clotted cream was as follows: : ·44,, •80 57.09 •57 Skim milk produced by abstracting cream from milk by means of the centrifugal cream separator was exceedingly poor in fat, containing generally less than 3, and only in exceptional cases more than 4 per cent. of fat. The butter samples were derived from three different sources; six referred to butter churned in the Aylesbury Dairy Company's dairy, fourteen to French, and fifteen to Danish and Swedish butter. I think it will be more instructive to give the results of the analyses of the three kinds separate. |