]\gt ; ' XVII .
On the Atomic Weight of .
T. S. HUMPIDGE , Ph. D. , B. Sc. Communicated by Professor E. Received lsIarch 20\mdash ; Read April 12 , 1883 .
[ PLATE 45 .
] I. Introductory .
EVER since the discovery of glucinum by VAUQUELIN , 1798 , its atomic weight has been a disputed matter amongst chemists .
Its discoverer considered that its oxide was a monoide , an opimon which was however strongly opposed by who wrote the oxide and the atomic weight .
The researches of and DEBRAy again turned the scale in favour of the earlier view , and as an atomic weight of suited the properties of the metal in the tables of periodicy constructed by MM .
and LOTHAR MEYER , this atomic weight has , up to quite recently , been generally accepted by chemists .
As a welcome confirmation to this came a determination of the specific heat of the metal by Professor E. REYNOLDS , who found that for its atomic heat to be near the normal number , its atomic weight must be and not .
Almost immediately afterwards a second determination of the specific heat was made by MM .
and PETTERSON , S who , however , obtained a result agreeing not with the lower atomic weight but with the higher .
The reasons for these conflicting opinions are to be found\mdash ; first , in the anomalous position of glucinum among the elements ; secondly , in the difficulties which surround the preparation of even small quantities of the free metal in a tolerably pure condition ; and thirdly , in the .ct that no volatile compound of glucinum is known of which the vapour density might be easily determined .
The constitution of the compounds ofglucinum cannot be inferred from any physical or chemical similarities with analogous compounds of other metals .
Its compounds most closely resemble those of magnesium and aluminium , but also differ these in the most striking manner .
None are isomorphous with any similar compounds of these .
Ann. , lvi .
, 101 .
Ann. Chim .
and Phys. [ 3 ] , xliv .
, 5 .
Phil. Mag. [ 5 ] , , 38 ; Chem. News , xlii .
, 27.3 , S Berl .
Ber .
xi .
, 381 , 906 .
drate iaustic stash , basic counds , luorides wotassium aodium , hetwo mxcept tbtained theoxide isomorphous wumina.oride , endency t num ; while its carbonates , its double sulphate with potassium , and the greater simplicity of some of its double compounds when its oxide is written as a monoxide show similarity with magnesium .
A correct solution of this long-vexed question can therefore be only obtained either from the specific heat of the element or from the vapour density of some of its volatile compounds .
Two tions of the specific heat have been made , as referred to above , but with quite contradictory results ; and , as far as I am aware , no determination of the vapour-density of glucinum chloride , nor of its compounds with ethyl and propyl , discovered by CAHOURS , has yet been made .
The metal used by Mr. NOLDS in his experiments was prepared in a platinum crucible , and was apparently impure .
His calorimeter , or as he prefers to it , tometer , was " " essentially a thermometer with a test-tube sealed in the bulb In the absence of any detailed description of his apparatus , and of the means adopted for its graduation , it cannot be definitely decided whether accurate results were possible or not .
M. NILSON used iron vessels to prepare his metal and carefully determined its composition .
It contained 13 per cent. of various impurities , which were avowed for in calculating the actual specific heat of the metal .
The determinations were made with BuNSEN'S accurate and delicate ice-calorimeter .
The researches detailed in this paper were undertaken to obtain , if possible , a purer metal than M. NILSON'S , and to redetermine its specific heat .
The vapour-density of volatile glucinum compounds I hope to refer to in a later paper .
II .
of Of the various methods recommended for the extraction of the earth from its commonest mineral beryl , I shall here ondy allude to two .
The first consists in decomposing the finely-powdered mineral by fusion with potassium carbonate , disintegration of the fused mass with sulphuric acid , evaporation to dryness , to render the silica insoluble , concentration to separate most of alumina , as alum , and treatment with ammonium carbonate as in the second method .
This process , otherwise a good one , is objectionable , because of the high temperature necessary to decompose the mineral , and because the fused mass is only very slowly attacked by the sulphuric acid .
I have usually followed a different method , due to SCHEFFER , which consists in decomposing the mineral by hydrofluoric acid .
Briefly , the process is as \mdash ; The finelypowdered mineral is mixed excess of powdered fluor spar and sulphuric acid * Ann. .
Ch. .
Pharm. , lxxx .
, 211 .
Jahres .
Ber .
1859 , 139 , completely precipitate with sodium sulphide .
This is the only method I know of , except fusion with acid potassium fluoride , which removes every trace of iron .
In the filtered solution the ammonium carbonate is now to be decomposed by boiling , but if the concentrated solution is heated directly over the flame very violent bumping cannot be avoided .
A better plan is either to dilute freely with water ( to about three times the bulk ) , or else to warm for about two houls over a and when most of the ammonium carbonate has been decomposed , then to boil over the flame .
The final traces of the rbonate are removed by acidulating with hydrochloric acid , yielding a solution from which ammonia precipitates the hydrate .
By repeating the process the glucina is obtained nearly pure .
Absolutely pure it can only be prepared from the double potassium fluoride , , by fusion with acid potassium fluoride , and recrystallizing the double fluoride from hot watel .
The quantity of the acid fluoride required by theory ( 6 : 1 ) should be used , not only 2:1 , as recommended by Dr. WOLCOTT GIBBS , who , I believe , first introduced the process .
The purified double fluoride is then readily decomposed by the requisite quantity of concentrated sulphuric acid in a platinum dish , and the hydrate precipitated with ammonia in the usual manner .
A very good yield may be obtained from beryl by this hydrofluoric acid process , provided the mineral is sufficiently finely powdered .
Fairly pure of glucinum may , however , be now obtained from some continental manufacturing chemists , which only require once purifying with ammonium carbonate for all ordinary purposes .
III .
Separation of glucinum .
Metallic glucinum may be obtained either from its double fluoride with potassium or from its chloride , of which the latter is decidedly the most suitable compound .
The double fluoride , , melts at a low red-heat without decomposition .
a considerable difficulty in excluding the oxygen of the air ; the tubes could scarcely be unscrewed again when cold , as the sodium chloride had solidified in the thread , and when opened it was not easy to extract the metal without contamination from the outside scale .
A modification of the two methods was therefore adopted , which answers admirably .
Iron boats were constructed of pieces of tubing partly filed away , and with end pieces screwed on .
They were about 200 millims. long and 20 millims. in diameter .
In order to prevent any contact of the reduced metal with the glass of the tubes , the boats were enclosed in slightly longer pieces of entire iron tubing , and in these cases were introduced into a wide tube of hard glass .
As in DEBRAY'S method , the first boat contained about the requisite quantity of sodium , the second the glucinum chloride .
When the tube had been filled with dry hydrogen , the boat containing the sodium was heated until the metal was fused , the heating was then continued backwards to the chloride , the vapour of which thus passed , mixed with the excess of hydrogen , over the molten sodium .
After the somewhat violent reaction had ceased and the tube was again cold , the boat containing the reduced glucinum , together with the sodium chloride produced and the excess of sodium , was placed in a vessel of crude alcohol until hydrogen ceased to come off It was then removed to water to dissolve out the sodium chloride .
Compact crystalline masses of glucinum were thus obtained , always , however , mixed with more or less basic chloride and oxide , even when the experiment was most carefully conducted .
The larger pieces of metal could be easily removed from the liquid and purified by washing with cold dilute caustio soda , which dissolves out any oxide , but is without action on the metal .
The residue was then likewise treated with caustic soda , and preserved separately as being less pure .
The iron boats and tubes were not attacked during the reaction , except that blisters were raised on the outer tubes which , unlike the boats themselves , were of steel .
The metal was of a steel-grey colour not so white as aluminium , but considerably harder , and can be beaten into thin sheets under the hammer .
Compressed in a steel mortar it yielded a compact disc of gramme thus treated gave a disc 15 millims. in diameter and 3 millims. thick .
Its specihc gravity at C. was found to be , or making ffiowance for the impurities present , 1.70 .
As it was suspected that the metal so compressed might still contain air , it was first boiled for half an hour in water before determining its specific gravity .
All attempts to fuse the metal were without any good result .
It remained apparently unaltered when heated to bright redness in the air , and probably became covered with a thin coating of the oxide .
Heated under sodium chloride in a lime crucible by the oxyhydrogen flame imperfect fusion was obtained .
When similarly heated and exposed to the air it -burnt with a dazzling , bluish light .
The metal dissolved readily in dilute acids leaving a minute IDCCCLXXXIII .
41 DR. T. S. HUMPIDGE ON THE ATOMIC WEIGHT OF GLUCINUM .
IV .
Deterrnination of specific heat .
The most accurate method for determining the specific heat of substances of which small quantities only are available is undoubtedly that devised by BUNSEN , in which , as is known , the quantity of ice at which is melted by a given weight of the substance heated to a high temperature is measured by the diminished volume which the water produced occupies .
Unless , however , the whole apparatus is kept exactly at the freezing-point , accurate results are not possible , and even undel the most favourable conditions there is usually a mean error of nearly one per cent. BuNSHiN originally immersed his instrument in a large box of pure freshly-fallen snow to preserve it accurately at proceeding which has been somewhat simplified by SCHULLER and WARTHA , who coat the exterior vessel with a thick layer of ice , and work in ice-cold water .
But even with this modification it is scarcely possible to use the apparatus in England , and especially on the west coast , where frost in the winter is the exception rather than the rule .
Several kinds of rough calorimeters based on the same principle have been proposed in which the heated substance produces an expansion in a liquid at a definite temperature .
Among these are the atometer of Mr. REYNOLDS , which was used to determine the specific heat of his glucinum , and a similar form described by Professor BALFOUR STEWART .
* No instrument of this description can , however , give accurate results unless most carefully shielded from external sources of heat and carefully calibrated , especially when a liquid expanding so irregularly as alcohol is used .
As far as can be judged by the published results , no calibration has been attempted , and therefore the results obtained cannot be of much value .
A very simple form of apparatus has been devised by KOPP who places the substances in a glass tube with naphtha , and heats in a bath of mercury .
Considerable dexterity must be requisite to remove the heated substance from the bath of mercury to the calorimeter in exactly equal times , and a further objection is the small range of temperature through which the substances can be heated .
His results , too , do not agree so closely with one another as those obtained by BUNSEN'S or methods , especially when only small quantities are used .
To overcome these various objections a modification of REGNAULT'S method of mixtures has been adopted , which with small quantities of the substances to 7 grarnmes ) gives results with a mean error of one per cent. or less , and be easily worked by one person without assistance .
The errors in 's appaxatus have been pointed out by NEUMANN and PAPE .
They showed how a cooling efFect must be produced by an upward current of cold air when the heating arrangement was opened at the top , and explained that , in many cases , the water equivalent of the * Proceed .
of the Phys. Soc. , , 52 , 342 .
Phil. Trans. , 1865 , 71 .
POGG .
Ann. , , 337 , 579 .
DR. T. S. JIUMPIDGE ON THE ATOMIC WEIGHT OF GLUCINUM .
each side .
The wires from those solenoids terminate in four points of German silver at the end of the board ( only one of which , J , is shown in the figure ) , which when tho calorimeter is drawn up into its place under the heater come into contact with four springs of the same metal ( K ) connected with the battery and a switch .
The board on which the calorimeter nd its appurtenances are fixed can be moved by strings from the position shown in the figure to that exactly under the heater and back again .
* obtain accurate readings of the different temperatures , two standard thermo* meters were first constructed , with a millimeter scale .
and calibrated by GAY-LUSSAC'S method , using Mr. F. D. excellent little instrument .
These two read between about and C. , and 1 millim. is about equivalent to C. They were compared with one another for every half degree between and C. , in a large bucket of water , kept suitably agitated , and in only one instance differed by more than from one another .
The bent thermometer was more open , and 1 millim. was about equal to C. It was very carefully compared , by two series of experiments , with the two standard thermometers together , and the mean temperature taken when they differed from one another .
The thermometer , used for the heater , read accurately to C. , was corrected for the boiling point , but not calibrated ; the correction for the exposed thread was determined experimentally , and was found to be less than given by the usual formula:\mdash ; The water-equivalent of the thermometer was calculated , from the weight of mercury which it contained , and the weight of glass immersed , to be gramme , which agreed with experimental results .
Readings of the temperature to which the substance was heated are thus correct to at least C. , and of the calorimeter to C. The liquid used in the calorimeter is French turpentine , purified and redistilled .
Taking the specific heat of this liquid as , and its specific gravity as , about 28 small calories are requisite to raise the temperature of .
through one degree , and only a small weight of the substance is required .
Thus , for silver , the quantity necessary would be grammes ( circa ) , and for aluminium gramme ( circa ) for a fall in temperature of : much smaller weights than are required for any other accurate method , except BUNSEN'S .
With a temperature difference of , and reading to , the mean error cannot well be smaller than 1 per cent. , and with the instrument in its present form , this is , in fact , the limit of error .
More accurate results might be possible with a still smaller calorimeter , and a still more open thermometer .
One possible objection to the use of turpentine was loss by evaporation during the *A battery-power of four quart BUNSEN or GROVES ' cells is necessary .
More convenient than this is a bichromate battery arranged that the zincs can be lowered into the liquid when required , Phil. Mag. [ 5 ] , , 57 .
a few times and a reading made at the end of the half minute .
Renewed agitations and readings are then made for every half minute until the maximum temperature is reached .
The time to reach this maximum for metallic substances , with which I have as yet only experimented , is about a half to one minute .
Although it is not intended to use the calorimeter with water , as the temperature difference would be too small for the small quantities employed , three experiments were made with pure silver , and two with commercial aluminium ( re-fused under sodium chloride ) to test the apparatus .
The following results were obtained , in which and have the meanings given above , is the temperature of the air , the weight of water , together with the water equivalent of the calorimeter , agitator and thermometer ( grammes ) , the weight of the substance employed , and the required specific heat .
Silver in water . .
II .
III .
Mean specific heat ; mean error per cent. Alurninium in . .
Mean specific heat ; mean error per cent. It thus be evident that results comparable in accuracy with the besb determinations can be obtained with the apparatus , and , I need not add , working entirely without assistance .
I intend shortly to make a series of more extended experiments , * Using a wider exit tube .
taking the specific heat of iron as , and that of glucina as .
Multiplying this number by the atomic weight of glucinum when its oxide is a sesquioxide , i.e. , 1365 , the atomic heat becomes , proving conclusively , if DULONG and PETIT'S law is true for this metal as it is for all others , that this is the true atomic weight , and not two-thirds of this number , or The specific heat , as above determined , is considerably higher than that found by NILSON , and my results might be too great for two reasons .
The compressed metal was somewhat porous , and heat might have been produced by the absorption of the turpentine in its pores , and there might be a further error from hygroscopic water .
The former supposition could be only decided by enclosing t , he metal in some impervious envelope , which has not yet been done ; the Jatter possible error was as as practicable avoided by drying the glucinum for two hours at before each experiment .
But , even supposing the above lesults to be erroneous to the extent of 10 per cent. , it is still certain that the specific heat of the metal is nothing near as it ought to be if the atomic weight were The result is unfortunate for the periodic law , and is the first selious rebuff which this useful generalisation of facts has received .
Glucinum , with the atomic weight of 9 , finds its natural position as the first element of the magnesium group ; but if its atomic weight ] , it falls between carbon and nitrogen , amd entirely contradicts the first principles upon which the tables have been constructed .
With an atomic weight of less than 12 it might be easily introduced , and it is possible that when absolutely pure such a result might be obtained .
NILSON has recently made some determinations with the sulphate which reduced the atomic weight from the number previously given to , but it does not seem quite certain that his compounds were entirely free from aluminium , which would naturally increase the atomic weight .
I hope shortly to re-determine this with the purest material it is possible to obtain .
DCCCLXXXIII ,

