490 Prof. J. Dewar and A. Scott .
[ Dec. 11 , III .
" Further Experiments on the Vapour Densities of Potassium and Sodium .
" By Professor James Dewar , M.A. , F.R.S. , and Alexander Scott , B.A. Received November 27 , 1879 .
In onr former communication " On the Vapour Densities of Potassium and Sodium , " we pointed out the chief obstacles which lay in the way of an exact determination of these constants .
Having overcome the chief manipulative difficulties in connexion with the method we described , there still remained the problem for solution as to how far the use of iron bottles in our experiments might affect the results .
If the iron retained the metals or allowed their vapours to diffuse with rapidity through it , a considerable error might be produced without its being easily detected .
We noted an action of this kind , remarking that " both metals seemed to have a tendency to behave in the same manner as zinc , viz. , to form an alloy with the iron , because the vapour evolved after a few minutes was absorbed completely , in spite of all precautions to prevent oxidation .
" The tendency of iron and the alkali metals to form alloys , we thought , would be diminished , so as not to interfere with the value of the results , " by raising the temperature as much as possible , so as to induce rapid volatilisation , " and the fair accordance of the series of experiments formerly given led us to suppose that this difficulty had been success- , fully obviated .
Recent spectroscopic observations , by Professors Liveing and Dewar , on the luminous absorption of the vapours of sodium under different conditions in iron and platinum vessels\#151 ; which will be presented shortly to the Society\#151 ; having led to the inference that mere rapidity of volatilisation , of such small weights of the metals as can be conveniently employed for the measurement of the volumes of the vapours , in presence of a relatively enormous mass of iron , would not eliminate the absorptive action , new experiments were made in a platinum apparatus of similar shape and size to the iron one , and in the same manner as formerly described .
Magnesia was used instead of sand to pack the platinum vessel in the crucibles employed as a protection against the diffusion of furnace gases , and the volatilisation of the alkali metals took place in nitrogen .
Bach set of experiments was made at different times , and the results are given in the order of observation , the bottle being freed from the vapours after each experiment by a strong current of hydrogen .
The molecular weights are calculated for 22'34 cub. centime , at 0 ' C. and 760 mm. pressure .
The following tables contain the record of the results obtained:\#151 ; 1879.1 On the Vapour Densities of Potassium and Sodium .
491 Potassium .
I. '046 grm. gave. . . .
24'1 c.c. at 0 ' and 760 mm. Mol. wt. . .
42-6 '04\#163 ; .
231 yy yy \gt ; \gt ; ? .
43-5 .042 " ... .
21'2 yy yy ) ) . .
44 2 .045 .
22-9 yy yy yy . .
43-9 .045 .
221 yy yy yy .
45'5 '030 . .
251 yy yy yy .
44'5 Mean molecular weight. .
44'03 II .
*047 grm. gave ... .
26*8 c.c. at 0 ' and 760 mm. Mol. wt ... 39*2 .049 " ... .
26'0 " " " 42-1 Mean molecular weight. .
40'65 Sodium .
I. '032 grm. gave ... .
26'5 c.c. 0 ' and 760 mm. Mol. wt ... ... .27 0 .033 ---26'7 " \#187 ; \#187 ; \#187 ; " " 27'6 .031 * " ... .
27'9 " \#187 ; 24-8 Mean molecular weight ... .
26 4 II .
'032 grm. gave . . . .
25'1 c.c. 0 ' and 760 mm. Mol. wt. '032 \#171 ; 5 . . . .
26*3 yy * yy 55 '031 yy . . . .
27'9 yy yy 55 .029 yy . . . .
26'5 yy yy 55 '031 yy . . .
\#171 ; 26'9 yy yy 5 5 .030 yy . . .
' 27*3 yy yy 55 .030 yy . .
* * 261 yy - yy 55 .
28-5 .
27'2 .
24-8 .
24'4 .
25-7 .
24'6 .
25'6 Mean molecular weight. . . .
25'8 492 On Vapour Densities of Potassium and Sodium .
[ Dec. 1 1 , '027 grm. gave . .
III , . .
24'6 c.c. at 0 ' and 760 mm. Mol. wt ... , . .
24*5 .027 " . .
23-2 99 95 55 . . . .
26-0 .037 .
" - - . . . .
31-7 99 59 9 9 . .
\lt ; . .
26*1 .030 " . .
29T 95 59 99 . . . .
23 0 Mean molecular weight. . . .
24-9 The new determinations of the molecular weights of potassium and sodium are just about half the former values , and would seem to support the inference that the atom of each of these metals resembles mercury and cadmium in the gaseous state , as regards molecular volume .
Such a remarkable result cannot be accepted without a very thorough investigation of the secondary reactions which may be induced at high temperatures .
It is certain that in the platinum vessels there is much less absorption of vapour than in the case of iron , and that the action of the furnace gases is all but eliminated .
The following suggestions might be made to explain the results .
In the first place , the metals may contain a large amount of occluded hydrogen .
With reference to this assumption , we have examined the sodium used in our experiments and find , on careful exhaustion and boiling in a Sprengel vacuum , that one volume of sodium only contains one-third of its volume .
of occluded hydrogen , so that this will not account " for the double volume of the atom of the metals .
" The alkali metals may form volatile compounds with platinum , although , judging from analogy , such compounds are not likely to exist .
If so , the vapour blown out of the bottle ought to contain platinum .
The use of platinum tubes to blow out the vapours prevented the test being applied .
The metals may be changed into oxides from occluded oxygen in the platinum .
If so , no free metal ought to remain in the bottle after this rapid evaporation .
This however is not the case ; so that , unless some other explanation can be given , it will be necessary to admit some anomaly in the behaviour of potassium and sodium , as regards their vapour densities at high temperatures .
In future the variation of the density with the temperature must be determined , but this is very difficult at temperatures about 1,500 ' C. The chief drawback to the prosecution of the experiments is the terrible waste of platinum vessels , which never stand the combined action of potassium , sodium , and hydrogen at high temperatures , beyond two or three operations .
The following determinations of the molecular weight of potassium iodide , made in the iron apparatus , are of importance in connexion with the previous results .
1879 .
] On Most General Problems in Continuous Beams .
Potassium Iodide .
*159 grm. gave ... .
2075 at 0 ' and 760 mm. Mol. wt. . . . .
171 .165 55 * * * * 21-2 55 55 .\gt ; * .
174 '170 55 . . . .
21-4 " 55 55 . . . .
177 '180 55 . . . .
216 55 55 . . . .
186 '103 55 * * * * 15-1 55 55 . .
\#171 ; .
152 174 55 . . .
\#171 ; 24-4 " 55 55 * .
* .
159 Mean molecular weight. . . . .
169-8 These numbers are fairly accordant , and seem to indicate that iodide of potassium is normal in its behaviour .
The vapour , after each experiment , was blown out with a current of hydrogen .
A crystalline deposit was obtained in each case , which was pure iodide of potassium , free from any trace of iron .
Taken in connexion with the former experiments , this seems to show that , if free potassium is abnormal , its compounds are not altered .
Before any final conclusions can be reached , further experiments must be made in platinum or iridium vessels , and it will be very important to control the results by examining the densities of the iodides of caesium and rubidium .
IY .
" On the Practical Solution of the Most General Problems in Continuous Beams .
" By John Perry and W. E. Ayrton .
Communicated by Fleeming Jenkin , F.R.S.S. Lond. and Edin .
, Professor of Civil Engineering .
in the University ol Edinburgh .
Received Noveriiber 27 , 1879 .
1 .
It is not necessary to enter into the question of the advisability of employing continuous girders in bridges with spans of less than 200 feet , but it is generally conceded that the increased economy due to the employment of continuous girders in longer spans more than counterbalances the well-known practical objections to continuity .
Hence the practical solution of the general problem\#151 ; given the conditions at the ends of a continuous girder , the spans , the moment of inertia of all cross sections , and the loading , to find the bending moment and shearing stress in every cross-section , is not unworthy of our attention .
2 .
Mr. Heppel has published the exact solution for cases in which each span may be supposed divided into two , three , four or five equal parts , in each of which the load and cross-section are supposed to be constant .
As is well known , the difficulty of the general problem is due to the necessity of making certain summations in each span .
How , Mr.

