With the discovery of the remarkable power of gas absorption possessed by charcoal cooled to a low temperature (see below), it became possible to make these vessels of metal. Previously this could not be done with success, because gas occluded in the metal gradually escaped and vitiated the vacuum; but now any stray gas may be absorbed by means of charcoal so placed in a pocket within the vacuous space that it is cooled by the liquid in the interior of the vessel. Metal vacuum vessels (fig. 1), of a capacity of from 2 to 20 litres, may be formed of brass, copper, nickel or tinned iron, with necks of some alloy that is a bad conductor of heat, silvered glass vacuum cylinders being fitted as stoppers. Such flasks, when properly constructed, have an efficiency equal to that of the chemically-silvered glass vacuum vessels now commonly used in low temperature investigations, and they are obviously better adapted for transport. The principle of the Dewar vessel is utilized in the Thermos flasks which are now extensively manufactured and employed for keeping liquids warm in hospitals, &c.
Thermal Transparency at Low Temperatures.—The proposition, once enunciated by Pictet, that at low temperatures all substances have practically the same thermal transparency, and are equally ineffective as non-conductors of heat, is based on erroneous observations. It is true that if the space between the two walls of a double-walled vessel is packed with substances like carbon, magnesia, or silica, liquid air placed in the interior will boil off even more quickly than it will when the space merely contains air at atmospheric pressure; but in such cases it is not so much the carbon, &c., that bring about the transference of heat, as the air contained in their interstices. If this air be pumped out such substances are seen to exert a very considerable influence in stopping the influx of heat, and a vacuum vessel which has the space between its two walls filled with a non-conducting material of this kind preserves a liquid gas even better than one in which that space is simply exhausted of air. In experiments on this point double-walled glass tubes, as nearly identical in shape and size as possible, were mounted in sets of three on a common stem which communicated with an air-pump, so that the degree of exhaustion in each was equal. In two of each three the space between the double walls was filled with the powdered material it was desired to test, the third being left empty and used as the standard. The time required for a certain quantity of liquid air to evaporate from the interior of this empty bulb being called 1, in each of the eight sets of triple tubes, the times required for the same quantity to boil off from the other pairs of tubes were as follows:—
| Charcoal | 5 | Lampblack | 5 |
| Magnesia | 2 | Silica | 4 |
| Graphite | 1.3 | Lampblack | 4 |
| Alumina | 3.3 | Lycopodium | 2.5 |
| Calcium carbonate | 2.5 | Barium carbonate | 1.3 |
| Calcium fluoride | 1.25 | Calcium phosphate | 2.7 |
| Phosphorus (amorphous) | 1 | Lead oxide | 2 |
| Mercuric iodide | 1.5 | Bismuth oxide | 6 |
Other experiments of the same kind made—(a) with similar vacuum vessels, but with the powders replaced by metallic and other septa; and (b) with vacuum vessels having their walls silvered, yielded the following results:—
| (a) | Vacuum space empty | 1 |
| Three turns silver paper, bright surface inside | 4 | |
| Three turns silver paper, bright surfaceoutside | 4 | |
| Vacuum space empty | 1 | |
| Three turns black paper, black outside | 3 | |
| Three turns black paper, black inside | 3 | |
| Vacuum space empty | 1 | |
| Three turns gold paper, gold outside | 4 | |
| Some pieces of goldleaf put in so as to make contact between walls of vacuum-tube | 0.3 | |
| Vacuum space empty | 1 | |
| Three turns, not touching, of sheet lead | 4 | |
| Three turns, not touching, of sheet aluminium | 4 | |
| (b) | Vacuum space empty, silvered on inside surfaces | 1 |
| Silica in silvered vacuum space | 1.1 | |
| Empty silvered vacuum | 1 | |
| Charcoal in silvered vacuum | 1.25 |
It appears from these experiments that silica, charcoal, lampblack, and oxide of bismuth all increase the heat insulations to four, five and six times that of the empty vacuum space. As the chief communication of heat through an exhausted space is by molecular bombardment, the fine powders must shorten the free path of the gaseous molecules, and the slow conduction of heat through the porous mass must make the conveyance of heat-energy more difficult than when the gas molecules can impinge upon the relatively hot outer glass surface, and then directly on the cold one without interruption. (See Proc. Roy. Inst. xv. 821-826.)
Density of Solids and Coefficients of Expansion at Low Temperatures.—The facility with which liquid gases, like oxygen or nitrogen, can be guarded from evaporation by the proper use of vacuum vessels (now called Dewar vessels), naturally suggests that the specific gravities of solid bodies can be got by direct weighing when immersed in such fluids. If the density of the liquid gas is accurately known, then the loss of weight by fluid displacement gives the specific gravity compared to water. The metals and alloys, or substances that can be got in large crystals, are the easiest to manipulate. If the body is only to be had in small crystals, then it must be compressed under strong hydraulic pressure into coherent blocks weighing about 40 to 50 grammes. Such an amount of material gives a very accurate density of the body about the boiling point of air, and a similar density taken in a suitable liquid at the ordinary temperature enables the mean coefficient of expansion between +15° C. and −185° C. to be determined. One of the most interesting results is that the density of ice at the boiling point of air is not more than 0.93, the mean coefficient of expansion being therefore 0.000081. As the value of the same coefficient between 0° C. and −27° C. is 0.000155, it is clear the rate of contraction is diminished to about one-half of what it was above the melting point of the ice. This suggests that by no possible cooling at our command is it likely we could ever make ice as dense as water at 0° C., far less 4° C. In other words, the volume of ice at the zero of temperature would not be the minimum volume of the water molecule, though we have every reason to believe it would be so in the case of the majority of known substances. Another substance of special interest is solid carbonic acid. This body has a density of 1.53 at −78° C. and 1.633 at −185° C., thus giving a mean coefficient of expansion between these temperatures of 0.00057. This value is only about 1⁄6 of the coefficient of expansion of the liquid carbonic acid gas just above its melting point, but it is still much greater at the low temperature than that of highly expansive solids like sulphur, which at 40° C. has a value of 0.00019. The following table gives the densities at the temperature of boiling liquid air (−185° C.) and at ordinary temperatures (17° C.), together with the mean coefficient of expansion between those temperatures, in the case of a number of hydrated salts and other substances:
Table I.
| Density at −185° C. | Density at +17° C. | Mean coefficient of expansion between −185° C. and +17° C. | |
| Aluminium sulphate (18)* | 1.7194 | 1.6913 | 0.0000811 |
| Sodium biborate (10) | 1.7284 | 1.6937 | 0.0001000 |
| Calcium chloride (6) | 1.7187 | 1.6775 | 0.0001191 |
| Magnesium chloride (6) | 1.6039 | 1.5693 | 0.0001072 |
| Potash alum (24) | 1.6414 | 1.6144 | 0.0000813 |
| Chrome alum (24) | 1.7842 | 1.7669 | 0.0000478 |
| Sodium carbonate (10) | 1.4926 | 1.4460 | 0.0001563 |
| Sodium phosphate (12) | 1.5446 | 1.5200 | 0.0000787 |
| Sodium thiosulphate (5) | 1.7635 | 1.7290 | 0.0000969 |
| Potassium ferrocyanide (3) | 1.8988 | 1.8533 | 0.0001195 |
| Potassium ferricyanide | 1.8944 | 1.8109 | 0.0002244 |
| Sodium nitro-prusside (4) | 1.7196 | 1.6803 | 0.0001138 |
| Ammonium chloride | 1.5757 | 1.5188 | 0.0001820 |
| Oxalic acid (2) | 1.7024 | 1.6145 | 0.0002643 |
| Methyl oxalate | 1.5278 | 1.4260 | 0.0003482 |
| Paraffin | 0.9770 | 0.9103 | 0.0003567 |
| Naphthalene | 1.2355 | 1.1589 | 0.0003200 |
| Chloral hydrate | 1.9744 | 1.9151 | 0.0001482 |
| Urea | 1.3617 | 1.3190 | 0.0001579 |
| Iodoform | 4.4459 | 4.1955 | 0.0002930 |
| Iodine | 4.8943 | 4.6631 | 0.0002510 |
| Sulphur | 2.0989 | 2.0522 | 0.0001152 |
| Mercury | 14.382 | .. | 0.0000881** |
| Sodium | 1.0056 | 0.972 | 0.0001810 |
| Graphite (Cumberland) | 2.1302 | 2.0990 | 0.0000733 |
| * The figures within parentheses refer to the number of molecules of water of crystallization. | |||
| ** −189° to −38.85° C. | |||
It will be seen from this table that, with the exception of carbonate of soda and chrome alum, the hydrated salts have a coefficient of expansion that does not differ greatly from that of ice at low temperatures. Iodoform is a highly expansive body like iodine, and oxalate of methyl has nearly as great a coefficient as paraffin, which is a very expansive solid, as are naphthalene and oxalic acid. The coefficient of solid mercury is about half that of the liquid metal, while that of sodium is about the value of mercury at ordinary temperatures. Further details on the subject can be found in the Proc. Roy. Inst. (1895), and Proc. Roy. Soc. (1902).