]\gt ; P RO CEEDINGS OF THE On the Atomic of Glucinum yllium)\ldquo ; Second Paper .
B.y T. S. HUMPIDGE , Ph. D. , B.Sc. , Professor of Chemistry in the University College of Wales , Aberystwyth .
Communicated by E. FRANKLAND , F.R.S. Received , 1885 .
Read March 5 .
In a former communication which I had the honour of to the Royal Society , * I described a method of metallic glucinum and of determining its specific heat .
From my experiments I deduced the resuIt that the atomic weight of the metal must be in order to agree with Dulong and Petit 's rule .
It is well known that the position assigned to glucinum in th periodic arrangement of the elements requires an atomic weight , of two-thirds the above number , or approximately 9 , and that with the larger atomic weight it falls between carbon and nitrogen , and is entirely out of place .
Various criticisms were offered to explain this apparenbly anomalous result .
To tue gestion of Professor J. E. Reynolds that the pure metal would have a specific heat 50 per cent. .
than that of a sample containing 6 per cent. of impurities , I have already replied .
Another explanation was by Brauner , S who thought it possible that the specific heat of the metal mi increase with the temperature and thus with Dulong and Petit 's rule , at some higher temperature-interval than .
Brauner 's suggestion was based upon the position which glucinum should } ) the periodic arrangement and which would be similar to that of boroIl and carbon .
Some determinations of the specific he it of the up Prof. T. S. Humpidge .
to 30 made by Nilson and Petterson*confirmed this prediction to certain extent , but as this important question could not be considered as definitely decided , the following were undertaken .
Owing to the small amount of time at my disposal the experiments have absorbed the greater part of my leisure for the past fourteen months , and in the mean time my results have been partly forestalled by other investigators , but the importance of the subject justifies me in publishing what is little more than a confirmation of the work of lers .
My experiments may be divided into two groups , ( i ) those on the fipecific heat of pure glucinum at varying temperatures , and ( ii ) those the vapour-density of volatile glucinum compounds .
I. Specific Heaf of Metallic Glucinum at varying Temperatures .
For the determinations of the specific heat of glucinum at varying temperatures a fresh quantity of the was prepared from the pure chloride by the method which I have previously described The chloride was obtained from oxide which had been carefully purified with ammonium carbonate .
xperiments in which impure glucinn was fused with acid potassium fluoride , and the fused mass extracted with water containing hydrofluoric acid , proved that if is present much of it goes into soIution with the glucinum .
The usual method with ammonium carbonate gives better results , especially if the proceHs is several times repeated , and if at last an insufficient quantity of the solvent is used , so that some of the glucina remains undissolved with the lasb traces of the alumina .
The concentrated soIution of the double can then be easily decomposed by leading steam into it through a wide tube .
The dangerous bumping which always happens wheIl bhe concentrated solution is boiled is thus completely avoided .
In this way , after three or four purifications , it is easy to obtain a sample of glucina which when fused with acid otassium fluoride dissolves in water containing hydrofluoric acid without leaving the slightest trace .
A very convenient form of charcoal for mixing with the oxide to prepare the chloride is a kind of lamp-biack known in trade as gas-bla .
When burnt it leaves a much smaller quantity of ash than the best sugar charcoal , and can be mixed with the oxide much more easily and intimately tbaIt the latter .
Most of the metallic glucinum prepared from this pure chloride was in the form of thin , lustrous , highly crystalline laminae .
The purer portions were selected and compressed in a steel mortar to a compact , lusbrous cylinder 11 mm. long and 8 mm. in diameter .
* ' Berlin Ber xiii , The latter experiments were done before the former .
See Appendix A. On the Atomic Weight of An analysis of a portion of this metal after the specific heat determinations had been done was conducted as follows .
Since aluminium was absent in the oxide and could not have been introduced during the preparation of the metal , the only impurities to belooked for were silicon , iron , and glucinum oxide .
soIution of tho metal in a dilute acid , the silicon would mostly remain behind in the free state , but a part would be liberated as silicon hydride .
* The oxide would also remain undissolved as the metal had been heated up to 45 in the specific heat determinations .
The iron would , of course , dissolve .
mgrms .
of the metal was dissolved in dilute sulphuric acid , and left a minute , imponderable residue , which was at mgrm .
It remained floating in the liquid for some time , light coloured , and was probably glucina mixed with some ferric oxide .
The iron in solution was estimated by the depth of colour produced by ammonium sulphocyanate , just as by the operation as " " Nesslerising\ldquo ; minute quantities of ammonia are estimated .
I believe this method gives more accurate results for the estimation of very small quantities of iron than the usual process with potassium permanganate .
In using a dilute solution of this reagent a considerable excess must be added before the pink colour becomes visible , and the end point or amount of excess is very difficuIt to determine ; whereas with anlmonium sulphocyanate and under similar conditions it is very easy to match the two colours and to determine accurately a minute quantity of iron .
The iron in the above solution of glucinum sulphate was oxidised by a drop of weak bromine-water , the excess of bromine boiled off , and a given volume of ammonium sulphocyanate added in a Nessler cylinder .
To another similar cylinder , containing the same quantity of free acid and ammonium sulphocyanate and made up to the same volume , was added a solution of ferric sulphate containing mgrm .
iron per .
Of this solution .
was required , corresponding to .
iron .
The composition of the metal was therefore:\mdash ; in other words , it was almost pure .
Its purity also shown by difficulty with which it dissolved in dilute .
A small piece which had been left in dilute sulphuric acid ( 1 : 10 ) over night was partly dissolved in the , but solutionl took place l.apidly on warming .
* The silicon in aves in 6 Prof. T. S. Humpidge .
of cold water flows .
In this way the chamber ( B ) into which the calorimeter has to be run is kept quite cool .
A double tin screen ( not shown in the fi serves also to prevent the calorimeter from receiving extraneous heat .
The whole operation of transferring the substance from the heater to the calorimeter can be in less than two seconds , and the gain of heat by the calorlmeter during this period cannot be measured .
In a series of blank experiments it was found that when the calorimeter remained under the heater for 15 seconds , the thermometer rose .
The temperature of the substance in the heater was measured by three delicate thermometers from Geissler .
Their fixed points were determined in ice , water , naphthalene , and benzophenone , but they were not calibrated .
The gas supply for this was regnlated by three regulators made by Giroud , of Paris , burning 60 , 80 , and 1101itres of gas per houl\ldquo ; and giving temperatures of and 310o respectively .
With these regulators the temperature could be kept constant to about , provided there were no great tions in the gas pressure .
I have sadly felt the need of a good regulatol for high temperatures .
* The suspension of the substance ( S ) in the heater was by a small pair of tongs , made so that the jaws open when the al'msal'e closed .
The tongs with the substance are supported by a thin platinum wire which is attached , outside the heater , to a thread passing over pulleys and fastened to a catch ( K ) .
When this catch is released the tongs fall by their own weight until they reach the ring X ; this brings the arms closer together , and the substance drops into the calorimeter .
The connexions are so arranged that when the calorimeter is rnn under the heater , the lid of the calorimeter casing is first opened , then the slider under the heater , and finally the catch is released , and the substance falls .
As the calorimeler returns the two sliders are again closod .
The liquid used in the calorimeter was purified turpentine , of which the specific was determined as follows:\mdash ; By the method of mixtures , using metallic magnesium , of which the specific had been found to be .
This gave for the turpentine By Andrews ' the mean of two series of experime1lts gave By Pfaundler 's apparatusS ( Joule 's principle ) , in which equal weights of water and turpentine contained in two similar calorimeters are heated by two platinum wires of equal resistance , through which *I am much indebted to Professor L. Meyer for advice and assistame in the construction of these two heaters .
For details of these and the following experiments , see Appendix C. ' ' Ann. Chim .
et Phys [ 3 ] , , 92 .
S " " Wien .
Akad .
Ber Prof. T. S. Humpidge .
Th quantity of heat required to raise 1 gram of a substance to can be expressed by the equation\mdash ; is the true specific heat at , and and are constants .
this equation the mean specific heat between any two temperatures ( and t ) is .
The numerical values obtained for the three unknowns the experimental values of ; ; A comparison of the value for the mean specific heats calculated the above expression with those obtained by direct experiment shows that the two agree within the experimenta , errors , and that this , expression correctly represents the change in the specific heat of gluoinnm .
' , found .
, calculated .
The three constants being known , the true specific heat at any is found by making , and the above equation becomes or with numerical values , whence the following values for may be calculated\mdash ; Prof. T. S. Humpidge .
II .
Vapour-density of Volatile Glucinum Compounds .
The compounds of glucinum which can be volatilised unchanged are those which it with organic radicals and with the balogens .
Of the former class Cahours states*that he has prepared the ethide and propide , and gives their boiling-points , but all my efforts to prepare either of these compounds in a sufficiently pure state to deter- mine its vapour-density were fruitless .
I operated , as recommends , on an excess of metallic glucinum with mercuric ethide or propide in sealed tubes ; but either the tubes exploded violently , always when a temperature of was exceeded , or if the change successfully accomplished , as shown by the separation of metallic mercury , the compound decomposed again on distilling .
Mercuric ethide acts on glucinum chloride , slowly at , more rapidly at and a considerable quantity of mercury ethyl chloride is produced , which partly remains dissolved in the excess of mercuric ethide , and paxtly separates out in the usual pearly scales .
A similar reaction takes place between mercuric ethide and glucinum bromide , but from neither of these reactions couid a volatile compound of glucinum be btained .
I hope to be able to examine thess interesting reactions at a later date .
As the organic compounds of glucinum wers not available for vapour-density determinations , my attention was directed to the halogen compounds .
The bromide , chloride , and iodide of glucinum are all volatile , and their volatility is in the order named .
The first two volatilise without decomposition if water and air are absent ; the las-named decomposes on heating .
In the presence of a trace of both the chloride and bromids decompose when heated , the latter more easily than the former ; in both cases the haloid element is set free .
And since both these compounds corrode glass and porcelain when vaporised in vessels of these substances , the determination of their vapour-density is a matter of some difficulty .
I first made some determinations in glass tubes with the chloride , using Schwarz 's modification of .
Meyer 's displacement method .
The glass was corroded , and the results were not concordant .
Porcelain was also attacked .
Ths tubes neally always contained free chlorine , although the experiments were done in pure dry It was then decided to try platinum for the determination , and it had been found that glucinum chloride could be sublimed unchanged * ' ' Count .
rend lxxvi .
" " Berlin Ber , 1051 .
The obvious in this modification have been pointed out by V. Meyer Berlin Ber xvu , 1334 ) .
They are chiefly that a portion of the wide tube is ularly heated , and that the large boat which cal.ries the substanoe takes air with it in its fall into the tube .
On the Atomic of Glucinum .
in a platinum tube in a stream of nitrogen , a platinum vessel was obtained from Messrs. Johnson , Matthey , and Co. This platinum vessel was of the usual size and dimensions of those employed for density determinations by displaced , except that the narrow tube was about one-half the ordinary length .
The open end of this tnbe was firmly attached to the upper of the usual glass apparatus by well-wired caoutchouc tubing .
* The substance was oduced , and the displaced air measured by the method used by Meier and Crafts in their researches on the vapour-density of iodine .
As a graduabed vessel for the reception of the ) aced air , an ordinary calibrated burette was nsed , the lower end being connected with a tube of the same width by strong caoutchouc tubing .
These two tubes were held in common and clips , and could be easily adjusted , so that the mercury was at the same level in each .
They were connected with the vessel in furnace by a narrow glass tube ( 1 mm. bore ) , and tlIe one containing the was immersed in a vcsseI of water .
As a furnace , two of Fletcher 's draft furnaces were used , placed one above the other , with a wide clay cylinder between .
This gave a chamber 30 cm .
high and cm .
diameter , which could be heated up to any temperature from about to The tempel.ature could be kept sufficiently constant for these iments with a good gas tap .
The plalinum vessel was freely suspended in two clay crucibles , placed mouth to mouth , and with a hole knocked in the bottom of the upper .
This served to protect the vessel from the direct action of the furnace gases .
The temperature was measured by a Siemens ' pyrometer , the porcelain tube of which was connected with platinum wires to the platinum vessel .
With this apparatus the vapour-densities of chloride and bromide were determined .
Gluc.inum chloride was in several ways , but the best method was found to be that recommended by Nilson , which consists in heating a few .
of the metal in a stream of dry hydrochloric acid in a platinum tube .
The chloride is sublimed as near the end of the tube as possible , the connexions removed , and a small cap fitted on the open end .
The tube is en cut ofl with a pair of scissors , so as to form a small capsule enclosing the chloride .
This is introduced into a well-corked glass tube and weighed .
While the preparation of the chloride was going on , the * During the progress of this part of , I eceived the account of and Pettersson 's expel'iments on the vapour-density of glucinum chloride ( ' ' BerJin Ber xvii , 987 ) .
It will be observed from the sequel that I have adopted some details from them , the use of acid for displacemenl .
" " " " Berlin Ber , 861 .
For a description of the construction and tion of instrument , see Appendix B. Prof. T. S. Humpidge .
apparatus was gently heated , carbonic acid passed through it by a glass tube going to the bottom of the platinum vessel , until the issuing gas was completely absorbed by caustic potash .
* The glass tube was then gradually withdrawn , and the temperature raised .
As soon as the temperature had become constant , which was easily seen by the merculy remaining stationary in both tubes and the resistance of the pylometer being constant , the cork at the top of the vessel was removed for an instant , the small platinum tube with the substance dropped in , and cork immediately replaced .
The equilibrium was thus hardly disturbed , and a few minutes afterwards the substance was allowed to fall into the heated vessel .
In the two following determinations ( the only ones which were made ) evaporation commenced at once , and was complete in about twenty seconds .
The following are the daba of these two experiments:\mdash ; Exp. Substance .
Displaced 1 . . . . . .
c.c. fean After each experiment the platinum vessel was washed out with a mixture of potassium iodide and starch , but no trace of free chlorine was found .
The small inum tube was then washed with water , , and weighed ; this gave the weight of the substance used .
The vapour-density of glucinum bromide was next determined in the same apparatus .
This substance can be obtained like the chloride , by heating the metal in dry hydrobromic acid , but , owing to the difficulty of obtaining a regular stream of the dry gas , it is best prepared bulning the metal in gaseous bromine .
between the two takes place at a low red heat , with brilliant incarldescence .
crude bromide is afterwards purified by resubliming it in a current of hydrogen or carbonic .
Like the chloride , it can be sublimed unchanged in carbonic acid .
Glucinum bromide sublimes at a lower temperature than the chloride , and , unlike the latter compound , vaporises } ) efore it meIts .
It begins to volatilise at about When carefully sublimed at a low temperature , it forms beautiful snow-white silkyneedles , and does not attack glass if and moisture perfectly absent .
If heated in the presence of air , free bromine is at once produced .
A quantity of the crude substance was by healing a few *Ihere was always a minute bubble of air left unabsorbed by the potash , even when the gas was passed thrugh the apparatus for 20 ) .
It was due to diffusion through the caoutchouc connexions , and did not appear to in{luence the results .
On the Atomic of centigrams of motallic glucinum in bromine vapour .
This was purified by sublimation in dry hydrogen , and was finally snblimed into glass tubes , each of which was sealed up .
When a determination was to be made , the two ends of one of these tubes were cut off , and the solid lump of the bromide quickly rammed into a small platinum tube closed at one end .
A cap was fitted on this tube , and it was weighed in the same manner as the chloride .
The vapour-density was determined like that of the chloride , and the are the data of the four experiments made : \mdash ; In.these experiments there was a trace of free Lromine duced .
In experiment 4 , where it was largest , the iodine , set free on the addition of potassium iodide , was estimated by potassium al.senite , , of which c.c. was required .
This corresponds to mgrm .
bromine , or less than 1 per .
of the quantity present .
The decomposition which had taken place could not , therefore , affect the result to any appreciable extent , and that the substance had completely evaporated was proved by the fact that the small platinum tube weighed the same before and afber the experiment within one or two tenths of a milligram .
It is therefore evident that the molecules of these substances in t , he gaseous state are represented by the formulae and respectivoly , in which the metal is a dyad , and has the atomic weight .
Thus from 's law , to which there are no known exceptions , the conclusion is confirmed which was ined from the specific heat of the element at The long disputed question of the atomic of glucinum is thus definitely and finally decided in favour of that number satisfies the requirements of the periodic law , and nother elenlent added to the long list of those whose atomic have been corProf .
T. S. Humpidge .
rected by this important generalisation .
In all future determinations of the atomic weight of an element , the position which the element should occupy in the periodic arrangement must receive due importance , and had I ] recognised this two years ago , I should perhaps stated my conclusions and criticisms less positively than I did .
APPENDIX A. On the Composition of the Double bonate of Glucinum and Ammonium .
An ysis of this important compound has been published by Debray , the formula he obtained being somewhat complex , I have prepared and analysed the pure substance .
A strong solution of ammonium carbonate was digested with moist glucinum carbonate at a gentle heat until saturated .
The clear solution was then heated until it became cloudy , and any excess of ammonium rbonate decomposed , then again filtered and mixed with its own bulk of strong alcohol .
The crystals , which only separated slowly and adhered firmly to the walls of the containing vessel , were drain , , washed with ] alcohol , and dried between filter-paper .
The yses of two separate specimens gave the following results : \mdash ; which correspond to the formula This resembles the formula obtained by Debray , 3 , Gl but contains more water .
Both of my specimens were dry crystalline powders , without a trace of adbering alcohol , and with a faint odour of ammonia .
Specimen II had been kept for several weeks in a Htoppered bottle .
* ' ' Ann. .
Chim [ 3 ] , xliv , 5 .
On the of Note on } ometer .
The following are the details of construction and graduation of the pyrometer employed in the vapour-density determinations .
About 100 cm .
of thin pJatinum wire ( 013 mm. diameter ) was coiled ound a thick piece of ordinary clay pipe , and three thicker platinum wires were soldered with gold to two ends .
To these thicker wires were soldered three copper wires , which went to the bridge where the resistance was measured .
The whole was enclosed in a porcelain tube , one end of which had been closed by a plug of clay , and the , were so arranged that the junctions between the copper and platinum came at the same height , just within the .
tube .
By suitably arranging the three wires and a known resistance the arms of the bridge , measurements could bc made of the resistance of the thin coil that of the leads , so that any change in resistftnce was due to the altered resistance of the coil alone .
The wire of which the coil was made had previously been heated several times to redness .
In order to graduate the instrument it was immersed in melting ice , boiling water , aniline , diphen.ylamine , sulphur , seIenium , and zinc .
For boiling these substances ( with the exception of water ) iron tubcs welded together at one end , of 5 cm .
diameter and 46 cm .
le1tgth , were used .
The upper portion of the tube for about 8 cm .
was surrounded with a coil of thin lead tnbing , ' which a stream of cold water was allowed to flow .
This acted as a very efficient condenser , and kept the upper part of the tube cool to handle eusily .
The results obtained are expressed in the following table , in which the first column gives the substance , the second its melting or point , and the third the actual found .
In the fourth column the resistances are reduced to that at , and in the fifth are given the ding r istances , calculated from the cmpirical formula\mdash ; Prof. T. S. Humpidge .
The reement between the calculated and observed results is good , except in the last two instances .
The selenium probably contained some sulphur , which would lower its boiling point , and only one experiment } made with zinc , as this temperature could only be reached in the furnace used under an exceptionally favourable gas pressure .
The change in the resistance of commercial platinum has beeIl investigated by the late Sir W. Siemens , M. Benoit , and others .
Siemens expressed his results in the form\mdash ; , and this can be into a general expression milar to that which I have used , .
:\mdash ; Benoit 's resuIts*can also be expressed by a similar formula .
But though there is a general agreement between these different results , the values of the coefficients vary considerably , and it is therefore always necessary in using this pyrometer to calibrate it for every vire .
I have even found that the chango of resistance of two pieces of from the same bobbin varied considerably .
M. Benoit has kindly furnished me with the following synopsis of val.ious results , which , although they difler much from one anobher , cain all be expressed by the general formula given above:\mdash ; .
rend .
, lxxvi , 342 .
The formula is incorrectly given in Wiedemann 's " " Electricitat the sign of second coefficient positive instead of negative .
One of the tempel.atures us in this research was hat of boiling cadmium , ] was taken from lier and incorrect determinations ( wibh an iodine thermomefer ) of Deville and Troost as instead of .
M. Benoit has , hcwever , informed me , in a private communication , that on introducing this colrection the general form of equation remains the same .
" " Wied .
Ann xxiv , 215 .
Prof. T. S. Humpidge .
Expt. 6. .
Expt. 7 . .
Mean : and By Andrews ' Calorifer .
quivalent of imeter and thermometer , weight of waler .
The mean of five gave whence equivalent of calorifer - : In two series of experiments , nsing turpentine instead of water , the following results were obtained:\mdash ; Series i.\mdash ; Weight of turpentine Series ii.\mdash ; Weight of turpentine ( experiments ) .
whence ( iii .
) By Method .
bfean of three experiments gave , compared with water at whence The general mean of the three sets of determinations is which was adopted , with a correction for the temperature , in calculatin the specihc heat of glucinum in the following experiments:\mdash ; II .
Heat of In Turpentine .
Weight of substance , weight of platinum casing water equivalent of calorimeter , \amp ; c ,

