]\gt ; Earl of Berkeley and E. G. J. Hartley .
cyanide solution the membrane was less permeable to a new solute .
new observations on cane sugar ( and one or two other substances ) were therefore so arranged that each was preceded by either a definite experiment on a ferrocyanide or by merely putting pressure on a solution surrounding the membrane ; in either case the apparatus thoroughly washed out before it was filled with the solution about to be exalnined .
( 2 ) A number of observations on " " dynamic\ldquo ; osmotic pressure*are also recorded .
These were made for two reasons : first , there were hopes that we could correlate the " " solution leak\ldquo ; rate , that is , the amount of solute passing through the membrane in unit time , with the rate at which water passed through under a given pressure , and so obtain a correction ; and , secondly , we wished to make further observations on this method as applied to dilute solutions .
Unfortunately the connection between " " tion leak\ldquo ; and " " water rate\ldquo ; turns ont to be nor have we been able to find any other means of the effect of this leak ; there is , however , more promise in the second of the above reasons , as will be seen in the Appendix .
It may be mentioned that these experiments are uot yet complete .
For osmotic pressures under 25 atmospheres the special deadweight apparatus in ' Phil. Trans , vol. 209 , p. 319 , was used .
The density determinations were carried out in flasks with graduated necks , as mentioned in ' Phil. Trans vol. , p. 180 .
PART 1 .
The on Cane-Sugar Solutions.\mdash ; Most of the experiments were made on pure cane\ldquo ; sugar chased from the Army and Navy Stores and on " " mineral water crystals\ldquo ; provided by Messrs. Tate .
At a later date Messrs. Tate kindly gave us a supply of specially pure mineral water crystals ; we are glad to have an opportunity of thankmg them for this , and for the care they took in seeing that it reached us in a fresh and uncontaminated condition .
It may at once be stated that no difference could be detected in these three qualities by osmotic pressure methods ; further , some of the mineral water crystals were dissolved in water and precipitated by alcohol , the resulting product gave ( within the limits of uncertainty as to the concentration , see below ) the same osmotic pressure as before .
As a further check on possible unknown sources of error two samples of Kahlbaum 's purest sugar wefe examined .
See 'Roy .
Soc. Proc , vol. 82 , pp. 271-275 .
Earl of Berkeley and E. G. J. Hartley .
whilst the previous work on Oxford sugar gave atmospheres .
discrepancy is probably due to the presence of an impurity of.smaller molecular weight in the Oxford sugar ; we have no data for an exact estimate of its effect ; the fact that the observed osmotic pressures deviate from Boyle 's Law by about 100 per cent. would indicate that the influence of an impurity in these solutions would be greater than in more dilute ones .
Table III ives the final results ( these are plotted in Diagram 1 ) .
As already stated , we have not succeeded in getting a satisfactory correction for the " " solution leak We therefore tabulate the mean values ( giving double weight to those with no leak or only a trace ) of the experiments in which the solution leak was less than .
of Further Determinations of Direct Osmotic Pressures .
483 The range of our experiments overlaps Prof. Morse 's work , and his values , reduced to our weight concentration , we give for comparispn .
The reduction was obtained by plotting the numbers on a large scale , and by means of the steel ruler apparatus described by one of a curve was drawn through the points .
The osmotic pressures corresponding to our weight concentrations were then read off the graph .
It will be noticed that the two sets of observations do not agree , the deviation increasing as the osmotic pressure decreases , and at the lower values it appears to be greater than either of our experimental errors .
We are unable to suggest any adequate reason for this discrepancy , but it is noteworthy that the graph of our results when produced towards the origin passes much closer to that point than Prof. Morse 's .
In this connection , it may be mentioned that in 'Phil .
Trans , vol. 206 , p. 505 , we point out a disagreement between Prof. Morse and ourselves in the osmotic pressure of a solution containing 282 .
per litre .
We are pleased to find that in his latest published results Prof. Morse ( who curiously enough does not refer to our work ) has been able to bring this observation into line with ours ; possibly we may meet with similar good fortune in the cases now under discussion .
The Equilibrium Pressure of -Methyl Glucoside Sotutions.\mdash ; The substance was purchased from Messrs. Kahlbaum , and was purified by filtering the concentrated solution through a porcelain filter and repeatedly recrystallising .
During these operations care was taken that the solutions were never heated above C. The observations are given in Table Iy ( the best lts are plotted in Diagram 1 ) , where the columns are numbered as in Table I , and have the same significance .
The water content of the crystals was determined in the same way as with sugar .
'Phil .
Mag vol. 24 , pp. 668 ( 1912 ) .
Further Determinations of Direct Osvnotic Pressures .
485 Column ( 4 ) gives , as before , the " " solution leaks These are derived as follows : glucoside which had penetrated the membrane was given three days to difluse out of the porcelain into the water filling the tube , it was then washed out , evaporated to dryness on a water-bath , and weighed ; the process was repeated a second time and even a third time if judged .
In several cases the residue first obtained was ignited , and the ash , which was assumed to be derived from a trace of copper sulphate ( the tubes are always stored in very dilute copper sulphate solution ) , was weighed .
A determination of the non-volatile matter in the water was also made .
These two determinations afford a means of estimating the actual quantity of glucoside in the residue , and this is the quantity tabulated .
Attempts were made to determine the -methyl glucoside by means of 's solution ; although a satisfactory method was worked out for the pure substance in as dilute a solution as .
per cubic centimetre , yet when applied to the residues above mentioned , which give still more dilute solutions , the results were quite unreliable .
From an inspection of the Table it is obvious that the experiments are not so satisfactory as with cane sugar .
Equilibrium Osmotic Pressures of Isodulcite and -Tetramethyl Solutions.\mdash ; The isodulcite , purchased from Messrs. Kahlbaum , was purified by recrystallisation .
Some difficulty was experienced in determining the water content , as the crystals , when under .
atmospheric pressure , melt in their water of crystallisation to form a glass which cannot be dehydrated without decomposition .
The difficulty was avoided ( but the esults are not so satisfactory as with cane sugar ) by dehydrating in a vacuum .
The apparatus designed for this purpose is shown in the figure .
A stoppered ( stopper not shown ) weighing bulb , containing a known quantity of powdered cystals , fits into the lubricated grobnd joint in the jacket C. Mercury followed by sulphuric acid is run round the stem of , and the side tube is connected to an exhaust pump .
is sealed off when a good vacuum is reached , and the whole apparatus is placed in an oven .
Cold water is circulated through : the jacket and the temperature of the is raised cautiously to the Earl of Berkeley and E. G. J. Hartley .
melting point of the crystals ( between 70 and C. ) and held there most of the water is driven off ; further heating to about until the is constant is then out .
The tetramethyl ferrocyanide was prepared as already described by one u It was recrystallised from chloroform , the chloroform of crystallisation driven off by gentle heat and then recrystallised from water .
The water content ( the salt contains two molecules of water of.crystallisation ) was obtained by means of complete analyses of samples withdrawn from the powdered salt at the time of making up the solution .
The results for the two substances are tabulated in Table .
The headings to the columns have the same signification as before , and Column ( 8 ) gives the densities of the solution at C. As neither the isodulcite nor the methyl feYrocyanide are soluble enough to give large osmotic pressures , no further experiments were made .
PART 2 .
Table gives the results for the ferro- and examined .
columns ( 1 ) to ( 7 ) inclusive have the same significance as for Table I. Column ( 8 ) gives the value of the osmotic pressure as calculated from Boyle 's Law , on the assumption that the salts are not ionised .
Column ( 9 ) the densities of the solution at C. The weight concentrations in Column ( 1 ) are calculated for the anhydrous salt , and all cases the water content has been determined either directly or * E. G. J. Hartley , 'Journ .
Chem. Soc vol. 97 , pp. 172 , 1732 ( 1910 ) .
Earl of Berkeley and E. G. J. Hartley .
From the osmotic pressure results for the magnesium ferrocyanide , taken in conjunction with the fact that the salt is ionised , it would seem probable that some at least of the molecules are associated .
None of the other solutions give any indication of this .
Table gives the results of a further search for association in solution , together with details as to the solutions ( marked ) the osmotic pressures of which have already been published .
Column ( 1 ) ives the volume concentration , i.e. , the number of grammes of anhydrous salt per litre at C. ; in most cases the salts were analysed for their water content .
Columns ( 2 ) to ( 5 ) , inclusive , bear the same gnificance as in previous Tables .
Column ( 6 ) gives the Boyle 's Law osmotic pressure , Column ( 7 ) the densities of the solutions at C. In Table it will be noticed that the osmotic pressures of the following salts , magnesium sulphate , magnesium chromate and zinc sulphate , are all lower than Boyle 's Law .
Now all these salts are ionised , and we therefore be certain that some association has taken place .
If Compare RaouIt , ' Tonometrie ' ( Scientia Series ) , p. 90 .
Earl of Berkeley and E. G. J. Hartley .
magnesium silicofluoride is appreciably ionised it also can be added to list .
We have already shown that the ferrocyanides of calcium and strontium are associated ; tbus all the salts hitherto found to be are constituted with a dyad base and dyad acid radicle , and were it not the outstanding case of magnesium camphorate ( further experiments this salt at somewhat higher concentration would seem to be desirable ) might state that this points to a general law .
On the other hand , both the and the tetramebhyl ferrocyanides the latter see Appendix ) normal values for the molecular hence , if we may assume that the methyl radicle is equivalent to an metal , the evidence derived from the osmotic pressures does not confirm previous that potassium ferrocyanide is associated .
APPENDIX .
The experiments detailed below were made not only to try the but also for the purpose of determining the molecular weight of the what insoluble substance , tetramethyl ferrocyanide .
In the course of the experiments on the " " dynamic\ldquo ; osmotic pressures , was noticed that very often the water rate before and after putting equilibrium pressures on the solutions differed considerably ; possibly membrane alters during the experiment ; obviously an alteration in bility would account for our failure to obtain a satisfactory solution correction , it also affords an explanation of the poor results for the dynamic\ldquo ; osmotic pressures .
The following method gets over some of the difficulties , and seems to afford a convenient and rapid means of determining weights , or , if these be known , small osmotic pressures .
A solution of cane sugar is made up of about the same molecular tration as that of the substance to be examined , and both solutions at C. The osmotic pressure apparatus ( with the porcelain tube with water ) is set up in ice , and is so arranged that it can be without disturbance .
The sugar solution is poured in , and the rate at the level in the water-gauge falls is noted ; this rate , when corrected " " guard-ring leak , be regarded as the standard rate R. The then washed out of the apparatus and its place taken by the new 'Phil .
Trans , p. 336 ( 1909 ) .
See E. G. J. Hartley , " " A Case of Isomerism.in the Methylated ' Journ. Qhem .
Soc vol. 103 , p. 1196 ( 1913 ) .
This " " guard-ring leak\ldquo ; is mostly due to the passage of water through the caused by the hydrostatic pressure of the water in the gauge .
492 Further Determinations of Direct Osmotic essures .
and the new rate obtained .
Let and be the molecular weights the sugar and the unknown substance respectively , and and volume concentration .
Then A tabular statement of the two sets of experiments is given on All the rates are measured with the water-level as near the same height possible , and frequent measurements of the guard-ring leaks were Each measurement on the unknown solution is both preceded and by a measurement on cane-sugar , and for calculating the mean of values thus obtained has been used .
It should be mentioned that the rates do not come to a steady state until about half an hour has elapsed from the time of filling .
The numbers given are the mean of three consecutive values , and represent millimetres ( on graduated water-tube ) moved over in 15 minutes .
The volume of 1 cm .
the tube is The accuracy of the above is about 2 per cent. , but we think the experimental error would be reduced if the apparatus were designed specially the work ; and there seems to be no reason why still more dilute solutions* should not give equally good results , in which case this manner of determining molecular weights would be more expeditious than any other known method , not excepting that of freezing-points .
lt may not be out of place to point out that the solution of cane-sugar of .
per litre has a -point lowering of C. The accuracy of 2 per cent. on this means an estimation of the temperature to 1/ 1000o .
With the best Beckman thermometers is but an approximation , for it is not generally realised that there are corrections of this magnitude due to column , change in barometer , other causes .
* In " " dynamic\ldquo ; osmotic pressures Proc we were able to get very fair results with solutions containing as little as 2 .
of sugar to the litre .

