1891.] on Scientific Uses of Liquid Air. 395 



Instead of silvering the interior and exterior of the vacuum 

 vessels, it is found convenient when using mercury vacua to leave a 

 little excess of liquid mercury, in order that the act of filling the 

 inner vessel with liquid air should cause a fine silvery deposit of the 

 metal over the exterior surface of the inner vessel. In such a vessel 

 liquid air or oxygen shows no signs of ebullition, the surface remains 

 as quiet and still as if it was ordinary water. The supply of heat is 

 cut down to less than four per cent, of what it is without exhaustion 

 and silvering in good vacuum vessels. The result is that volatile 

 liquids can be kept thirty times longer. Such vessels do not, however, 

 maintain indefinitely the high standard of heat isolation they possess 

 the first time they are used. After repeated use all vacuum vessels 

 employed in the storage and manipulation of liquid air deteriorate. 

 Illustrations of the appearance of such vessels are given in Figs. 1 

 and 2. The rapidity with which a space is saturated with mercury 

 vapour (which we know exerts a pressure of about one-millionth of an 

 atmosphere) is easily proved by simply filling a barometer in the 

 usual way, and then instantly applying a sponge of liquid air to a 

 portion of the glass surface of the Torricellian vacuum space, when a 

 mercury mirror immediately deposits, it is important to know the 

 amount of mercury deposited from a saturated atmosphere which is 

 maintained (containing excess of liquid mercury) at the ordinary tem- 

 perature, the condensation taking place when liquid air or oxygen 

 is discharged into a vessel surrounded by such a Torricellian vacuum. 

 If the deposit on the cooled bulb is allowed to take place for a given 

 time, the outer vessel can then be broken and the amount of mercury 

 which coated the bulb ascertained by weighing. Knowing the surface 

 of the cooled bulb, the amount deposited per unit of area can be 

 calculated. In this way it was found that in ten minutes 2 milligrams 

 of mercury per square centimetre of surface was deposited. Considering 

 that one-tenth of a milligram of mercury in the form of saturated 

 vapour at the ordinary temperature corresponds to the volume of 1 litre, 

 this proves that the equivalent weight of 20 litres had been condensed 

 in the space of ten minutes. This plan of cooling a portion of the 

 surface of a vessel by the application of a liquid air sponge, enables us 

 to test our conclusions as to the amount of matter present in certain 

 vacua. Here is a globe of the capacity of 1 litre. It has been filled 

 with, presumably, nothing but the vapour of mercury, by boiling under 

 exhaustion and subsequent removal of all excess of liquid. Such a 

 flask ought to contain mercury in the gaseous state that would weigh 

 rather less than one-tenth of a milligram, assuming the ordinary 

 gaseous laws extend to pressures of less than one-millionth of an atmo- 

 sphere. Now we know by electric deposition that one-tenth of a milli- 

 gram of gold can be made to cover one square centimetre of surface 

 with a fine metallic deposit. Considering the general similarity in the 

 properties of mercury and gold, we should therefore anticipate that if 

 all the mercury vapour could be frozen out of the litre flask it would 

 also form a mirror about one square centimetre in area. But after 

 one such mirror is deposited, the renewed application of a second 



