329 1880 .
] Revision of the Atomic Weight of Aluminium .
The numbers given are not the result of one reading , but the mean of four or five .
The copper block , after being raised to the tempera ture at which the meniscus of the liquid disappeared , was allowed to cool till a cloud became visible , shortly afterwards followed by a reappearance of the meniscus .
The temperature was again raised , and so on , readings being taken each time .
They never differed by more than half a degree .
The tubes , therefore , must have possessed the temperature of the heater .
The experiments described in a former part of this paper , relating to the difference of refraction shown by a liquid above its so-called critical point , and the gas evolved from it , were repeated with carbon tetrachloride and carbon disulphide , and held good in both cases .
The phenomena observed differed in no particular from those already described .
In conclusion a few remarks on the liquefaction of the so-called permanent gases may not be deemed out of place .
If the deductions from the above experiments be correct , it follows that that form of matter which we call gas may be converted into liquid by pressure alone ; but the meniscus will never become visible , for the process of change is a gradual one .
To render the meniscus visible , it is necessary to take advantage of the fact that liquids under such circumstances have a much greater coefficient of expansion by heat , and conversely a much greater coefficient of contraction on withdrawal of heat , than gases .
It , therefore , becomes necessary to lower the temperature until the liquid by contraction acquires a specific gravity greater than that of its gas , and then , and not till then , does the phenomenon of a meniscus become observable .
III .
" Revision of the Atomic Weight of Aluminium .
" By J. W. Mallet , F.R.S. Received March 13 , 1880 .
( Abstract .
) After remarking that there is probably no one of the so-called chemical elements , equally abundant in nature with aluminium , and occurring in as numerous compounds , with regard to the atomic weight of which our knowledge has long rested upon so slender a foundation of accurate experiment , the author gives a brief statement of prior determinations , noticing in succession the researches of Berzelius , Sir Humphry Davy , Thomson , Mather , himself , Dumas , Charles Tissier , and Terreil , and quotes the numbers given in several of the more recent chemical handbooks , reports , \amp ; c. He then proceeds to describe a redetermination of this atomic 330 Mr. J.W. Mallet .
[ Apr weight , which occupied a large part of his leisure time during th last three years .
In this investigation he spared no pains toattak as precise a result as possible , and aimed especially at the discover^ and as far as possible removal , of sources of error connected with the methods employed .
The following general principles were ket\gt ; t in view :\#151 ; 1st .
That each process should he as simple as possible , and should involve as little as possible of known liability to error .
2nd .
That different and independent processes should be resorted to as the means of checking each other 's results , even though it may fairly be assumed that one is more advantageous than another .
3rd .
That each process should be carried out with quantities of material differing considerably from each other in successive experiments .
4th .
That only such other atomic weights should be involved as may be counted among those already known with the nearest approach to accuracy .
The most scrupulous care was taken in the purification and examination of all the reagents used , and , as far as possible , vessels of platinum or of hard porcelain were substituted for those of glass .
In the author 's first series of experiments , the alumina was determined which resulted from the ignition of a known weight of ammonium alum .
In the second series , a solution was made of a known weight of bromide of aluminiam , and was very exactly precipitated by a solution of nitrate of silver , and the weight of bromide of aluminium com-pared with that of the silver required for forming the solution of nitrate .
In the third series , bromide of aluminium was used for the preparation of pure aluminium , by reduction by sodium ; a carefully weighed portion of the pure aluminium was dissolved in a solution of caustic soda , and the hydrogen given off was estimated by volume ; or else burnt , and the weight of resulting water determined .
In the following epitome of the results , A , B , C denote groups of experiments under each series , the experiments of each group differing only in the quantity of material operated on , while the different groups differ in the particular preparation used , or in some feature of the method .
Thus in Series I , A was made with alum dried by exposure to air for two hours ; B with alum dried by exposure for twenty-four hours .
In Series II , A , B , C were made with successive portions of a distillate of bromide of aluminium .
In Series III , A was made by estimating the hydrogen by volume , B by estimating by weight the water resulting from its combustion .
Only the number of experiments in each group , the mean result from the group , and the probable error of the mean , are here given .
[ 880 .
] Revision of the Atomic Weight of Aluminium .
381 Series I. A. B. No. 5 5 Mean .
27*040 27*096 P. e. .
+ *0073 . .
+ *0054 Series II .
A. B. C. " No 3 ... ... 5 ... ... 3 li u. .
... . .
27*034 27*023 27*018 iyicdiL \#171 ; ... \#187 ; .
P. e. + .0049 +*0052 \#177 ; *0069 Series III .
A. B. No. 6 3 Mean 27*005 26*990 P. e. .
\#187 ; . . . . . .
+ *0033 c. . . .
+ *0046 In view of the gradual loss of water which , crystallised ammonium alum has been shown to undergo on exposure to the air , the author considers that of these various sets of experiments , Series I , B , is entitled to least confidence .
He is inclined to attach most weight to Series III , A , since the method used was very simple in principle , the determination of one of the two quantities concerned was rendered very exact by the great volume occupied by the hydrogen , the comparison was made directly with the standard element in our system of atomic weights , and not through the intervention of any other substance whose atomic weight must be assumed , and the agreement .of the results among themselves is particularly good .
The general mean from all the thirty experiments , if all be included , is Al=27*032 , with a probable error for this mean of +'0045 .
If Series I , B , be excluded , the mean of all the remaining twenty-five \#166 ; experiments is Al=27'019 , with a probable error of +'0030 .
The general result adds , the author hopes , aluminium to the , unfortunately still limited , list of those elementary substances whose atomic weights have been determined within the limits of precision attainable with our present means of experiment .
This result also adds one to the .
cases already on record of the numbers representing carefully determined atomic weights approaching closely to integers , and leads the author to say a word on the reconsideration of " Prout 's Law .
" Taking the following 18 elements as the only ones of which the atomic weights may be fairly considered as determined , with reference to hydrogen , with the greatest attainable precision , or a near approach thereto , namely oxygen , nitrogen , chlorine , VOL. XXX .
2 A Dr. De La Rue and H. W. Muller .
[ Apr. 22 , bromine , iodine , sulphur , potassium , sodium , lithium , silver , thallium aluminium , carbon , phosphorus , barium , calcium , magnesium , and lead , and making a reasonable allowance for the errors of the determinations , he calculates the probability that nine of those numbers should lie , as they are found to do , within 01 of integers , supposing the value of the true numbers to be determined by chance , and finds it only as 1 to 235 2 .
The exact figure for the chance will of course depend upon the limit of error taken ; but the above example seems sufficient to show that not only is Prout 's law not as yet absolutely overturned , but that a heavy and apparently increasing weight of probability in its favour , or in favour of some modification of it , exists and demands consideration .
IV .
" On the Height of the Aurora Borealis .
" By Warren De La Rue , M.A. , D.C.L. , F.R.S. , and Hugo W. Muller , Ph. D. , F.R.S. Received April 22 , 1880 .
pU5 ?
)vol .
171 , Our experiments on the electric discharge , which have been already published in the " Phil. Trans. " and the " Proceedings of the Royal Society , " enable us to state with some degree of probability the height of the Aurora Borealis , when its display is of maximum brilliancy , and also the height at which this phenomenon could not occur on account of the great tenuity of the atmosphere .
In Part III of our electric researches , " Phil. Trans. , ' we have shown that the least resistance to the discharge in hydrogen is at a pressure of 0#642 millim. , 845 M ; after this degree of exhaustion has been reached , a further reduction of pressure rapidly increases the resistance .
When the exhaustion has reached 0'002 millim. , 3 M , the discharge only just passes with a potential of 11,000 chloride of silver cells ( 11,330 volts ) ; at the highest exhaust we have been able to* obtain ( and which we believe has not been surpassed ) , namely , 0*000055 millim. , 0'066 M , not only did 11,000 cells fail to produce a discharge , but even a 1-inch spark from an induction coil could not do so .
Although we have not experimentally determined the pressure of least resistance for air , we have ascertained that while the discharge occurs in hydrogen at atmospheric pressure between disks 0'22 inch distant , they require to be approached to 0T3 inch to allow the discharge to take place in air .
We may therefore assume that the pressure of least resistance for air is 0*642 x 13 :0'379 millim. , 498 6 M. At a height of 37'67 miles above the sea level , the atmosphere would

