311 On the Filtration of Crystalloids and Colloids through Gelatine : icith special reference to the behaviour of Hcemolysins .
By J. A. Craw , British Medical Association Research Scholar .
( Communicated by Leonard Hill , M.B. , F.R.S. Received December 1 , 1905 , \#151 ; Read February 1 , 1906 .
) The current controversy between Ehrlich ( 1898 , 1903 ) and Arrhenius and Madsen ( 1902,1904 ) on the physical chemistry of the neutralisation of toxins by their specific antitoxins led the author ( IY , 1905 ) to an investigation of the relations existing between the toxin for red blood corpuscles secreted by B. megatherium and its specific antiserum .
One of the methods adopted consisted in the filtration of mixtures of megatherium haemolysin and antilysin through gelatine .
The lysin was found to pass into and even through the gelatine , whereas the antilysin was retained , and by means of the delicate blood test for free lysin it was possible to demonstrate that the two substances , on mixing in any proportions , do not completely neutralise each other .
These observations indicated that Ehrlich 's views on the toxin-antitoxin reaction required considerable modification , but a closer investigation showed that the hypothesis advanced by Arrhenius and Madsen agreed even less with the experimental facts .
On the other hand , the results were in entire harmony with the views advanced by Landsteiner ( 1903 ) and Bordet ( 1903 ) , which have been supported by Nernst ( 1904 ) and Craw ( 1905 ) that the toxin is adsorbed by the antitoxin much as a dye is by a tissue .
As this conclusion may considerably modify current ideas on the nature of the reaction and the constitution of toxins in general , it seemed advisable to inquire further into the physical chemistry of filtration through gelatine .
The present communication contains data of the filtration of various crystalloidal and colloidal solutions , including megatherium .
lysin , through various percentages of gelatine , under constant and variable pressures .
The work was partly carried out as Research Student at the Lister Institute of Preventive Medicine and was completed at the London Hospital Medical College by the aid of a scholarship from the British Medical Association .
Previous Work on the Gelatine Filter.\#151 ; The gelatine filter was introduced by C. J. Martin ( 1896 ) , and consists of a Pasteur-Chamberland candle , the pores of which are filled with solid gelatine .
The filter is fitted into a gunmetal jacket or filter case , which serves to hold the liquid to be filtered , and the upper end of the closed filter case is connected with a supply of air at a pressure of 30 to 100 atmospheres , which is used to force the liquid VOL. lxxvii.\#151 ; b. 2 A Mr. J. A. Craw .
On the Filtration of [ Dec. 1 through the gelatine .
From his observations with " wet " gelatine filters , i.e. , filters containing gelatine from which the normal content of water had not been removed by drying in air or otherwise , Martin* concluded that gelatine was impermeable to colloidal substances such as globulin , albumin , glycogen , and soluble starch , but partially permeable to albumoses and dextrin , and completely permeable to solutions of crystalloids , urea and dextrose .
It seemed , then , as if the gelatine filter was an instrument destined to play an important part in the investigation of physiological fluids .
Martin and Cherry ( 1898 ) applied the filter to the investigation of the course of the reaction occurring between diphtheria toxin and antitoxin , and likewise to the reaction of snake venom with antivenene , the toxin and venom being filter-passers , whereas the anti-bodiesf were retained .
From these experiments it seemed as if the toxin was completely neutralised by the antitoxin , but further investigation of the mechanism of gelatine filtration shows that no such absolute conclusion can be drawn .
E. Waymouth Eeid ( 1901 ) showed conclusively that crystalloids do not pass " wet " gelatine filters in unaltered concentration , and that although filters which had been dried to constant weight in dry air allowed certain crystalloidal solutions to pass unchanged , yet dextrose and sodium oleatej were partially retained .
Further , he found that the filtrate from serum S had not the same composition as the proteid-free serum , and that the residual fluid left in the filter case had a much higher concentration of organic substances of non-proteid character than either the original serum or the filtrate .
The Martin filter is not , therefore , a simple means of separating crystalloids from colloids .
One must not , however , under-estimate its value as an instrument for the analysis of physiological fluids , for although the filter shows considerable differences in permeability to various crystalloidal substances , I find ( IV , 1905 ) that these and inferior colloids are , on the whole , retained to a small extent compared with typical colloids .
The partial retention of filter-passers has an important bearing on the conclusions to be drawn from gelatine filtrations , for , if the concentration of the filter-passer to be tested for be small in the original fluid introduced into the filter , the gelatine may retain practically the whole amount and the filtrate contain only a quantity below the experimental error of observation .
This was found to be the case for neutral mixtures of megatherium lysin and antilysin , and mixtures * * * S * Loe .
dt .
t Cf .
also Brodie ( 1897 ) ( 1900 ) .
| Krafft ( 1902 ) considers soap solutions , such as sodium oleate , to be colloidal .
S Cf .
also Starling ( 1899 ) .
1905 .
] Crystalloids and Colloids through Gelatine .
containing excess of antilysin , by the author ; * the filtrates showed no haemolytic power , whereas the gelatine had stored up a considerable amount of free megatherium lysin .
In this light Martin and Cherry 's observations on diphtheria toxin and snake venom , mentioned above , are in entire agreement with my results for megatherium lysin , the free toxin and venom of the neutral mixtures being probably stored up in the gelatine of their filters .
On " Wet " and " Dry " Gelatine Filters.\#151 ; Martin and others used the filters in the " wet " state , i.e. , shortly after solidification of the gelatine in the pores of the candle , and , therefore , containing a considerable quantity of water , part of which on filtration will pass into the filtrate and so dilute the substance filtered .
To get rid of this difficulty , E. W. Eeid removed part of the water by drying Martin candles to constant weight in a current of dry air .
It seemed to me , however , that by drying the gelatine another difficulty might be introduced , viz. , a change in the size of the pores , which would render observations with " dry " filters of doubtful value .
Method.\#151 ; The rate of filtration of water through freshly prepared wet filters was observed during one hour and compared with the rates during a similar period , of filters which had been partially dried by standing in dry air for 10 hours , 24 hours , and 3 days .
Rates of Filtration.\#151 ; The freshly prepared wet filters gave fairly uniform rates , and allowed from 05 to 2 c.c. of water to pass per minute at 100 atmospheres pressure , the concentration of the gelatine in the pores being 9 per cent , and the temperature 10 ' to 15 ' C. Under the same .conditions the partially dried filters allowed from 5 to 20 c.c. to pass per minute at the beginning of the filtration , but the rate rapidly decreased .
On reducing the applied pressure to about one atmosphere , and allowing the water to flow through a partially dried filter at the rate of about 1 to 2 c.c. per minute , the filter gradually tightened , so much that after the passage of 30 to 50 c.c. a pressure of 100 atmospheres was necessary to maintain the same rate .
The longer the filters were dried the more marked was the porosity and the greater the amount of fluid which had to be pressed through before the filter tightened .
Conclusions.\#151 ; On partially drying gelatine filters the gelatine shrinks , and air passages are produced of greater diameter than the water passages of a wet filter .
The wide dry filter passages at first offer a free passage to the fluid filtered , and no material change is to be expected in the percentage composition of the latter .
The filter gradually changes in character until ultimately , a wet filter is obtained with pores of similar dimensions to those of a freshly prepared wet filter .
The various fractions of filtrate from a * Loc .
dt .
Mr. J. A. Craw .
On the Filtration of [ Dec. 1 , " dried " filter are not subject to the same conditions of filtration and are not , therefore , comparable .
Wet filters , on the other hand , show a much greater constancy in rate of filtration , and are , therefore , to be preferred .
On the Preparation of the Filters.\#151 ; To overcome the difficulty of dilution of the filtrates by the water of wet filters , the gelatine to be used in filling the candles was in several cases dissolved in the fluid to be filtered .
Thus the filtration of 0*8 per cent , sodium chloride and of megatherium lysin took place through 9 per cent , gelatine which had been dissolved in 08 per cent , sodium chloride , and T54 per cent , potassium iodide was filtered through 11 per cent , gelatine in 1*54 per cent , iodide .
A Pasteur-Chamberland candle of size B , which had in the majority of cases been heated for 10 hours at least in a muffle furnace to remove organic matter , was fixed into a brass socket with " Faraday " cement and fitted concentrically into the internally tinned gunmetal jacket or " filter case .
" After thoroughly washing through with about 400 c.c. of hot water and 250 c.c. of the solution of French gelatine to be used in forming the filter , at a temperature of 37 ' C. , the gelatine at 30 ' .C .
was slowly passed through the cooling filter at an air pressure just sufficient to cause about 1*5 c.c. or 15 drops to filter per minute .
The filtration was continued until no trace of air bubbles was visible in the drops , after which a further 50 c.c. were allowed to pass .
During the final filtration the upper surface of the gelatine solution in the filter case was not allowed to descend below a level 3 cm .
above the crown of the candle .
The filter was then removed , drained , placed in the neck of a flask containing a little water , and kept in the ice chest at 6 ' to 8 ' C. until required .
If the filter so prepared had an obviously thick skin of gelatine upon its surface , this was removed by immersing in the gelatine solution at 30 ' C. and draining as before .
In the comparative experiments with different concentrations of gelatine , the filters were treated in the manner last described to get membranes of approximately equal thickness .
Only those filters were used which allowed less than 2 c.c. or about 20 drops per minute to pass at 100 atmospheres pressure and 10 ' to 15 ' C. , and the majority of filters allowed only about 0'5 c.c. per minute to pass under these conditions .
After use the candles could be washed out with water at 50 ' C. , dried , heated in a muffle furnace , and on refilling with gelatine again gave reliable filters .
Filtration of a Typical Crystalloid , Sodium Chloride , through 9 per cent. Gelatine .
Sodium chloride was chosen on account of its characteristic crystalloidal properties and the important rdle it plays in physiological fluids .
Further , 1905 .
] Crystalloids and Colloids through Gelatine .
315 in a previous communication * it was found that the filtrates from megatherium lysin were strongly haemolytic , and it was necessary to determine if this haemolysis was due in any degree to change in the concentration of sodium chloride .
Method.\#151 ; The filters used contained 9 per cent , gelatine in 0-8 per cent. NaCl , and about 50 c.c. of 0*8 per cent. NaCl were pressed through at 100 atmospheres and 10 ' to 12 ' C. , the filtrates being collected in fractions of about 4 c.c. The NaCl content of 1 c.c. of the fractions of filtrate , the residual liquid left in the filter case and the original fluid , was determined by titration with 1/ 100 normal silver nitrate , using potassium chromate as an indicator .
The haemolytic powers of the same fluids were determined in these and all other experiments in this paper , unless otherwise stated , by mixing 1 c.c. of the respective fractions with 2 c.c. of 2'5 per cent , guinea-pig 's red corpuscles which had been washed and suspended in 0'8 per cent. NaCl .
The mixtures were heated to 37 ' C. for three hours , well shaken every 30 minutes , and allowed to sediment generally about 12 hours in the ice chest at 6 ' to 8 ' C. The intensity of tint of the supernatant fluid was then determined by comparison with the scale of a von Fleischl haemoglobinometer which had been standardised by a blood solution of known content .
Complete hsemolysis is indicated by the index ( 100 ) .
Examination for Sodium Chloride and Power.\#151 ; Experiment No. 2 , Table I , represents one out of four similar experiments .
Original Fluid : 0'81 per cent. NaCl given in the table as Orig .
( 100 ) .
Filtrates : The relative concentrations of the 1st , 4th , and 14th fractions were ( 47'1 ) , ( 89'5 ) , and ( 99'2 ) respectively .
Eesidual Fluid : When tested immediately after decompression had the value ( 100 ) exactly , but , on allowing to stand 12 hours in the filter case , showed the relative concentration ( 105 ) .
Hsemolytic Powers : No hsemolysis was obtained under the standard conditions with either filtrates , residual fluid , or the original saline .
Conclusions.\#151 ; The typical crystalloid , sodium chloride , is markedly retained on filtration through gelatine which originally contained the same concentration of saline as the fluid filtered .
The salt taken up by the gelatine is expressed or diffuses into the residual fluid after decompression .
The diminution in salt concentration in the filtrate is insufficient to cause hsemolysis under the standard conditions .
* Loc .
cit. Mr. J. A. Craw .
On the Filtration of [ Dec. 1 , Continued Filtration of Sodium Chloride through 9 'per cent. Gelatine .
As the last fraction of filtrate in No. 2 did not quite reach ( 100 ) , it seemed desirable to ascertain whether on continued filtration this value could be obtained .
Method.\#151 ; The residual fluid of No. 2 , which had stood in the filter case Table I. Sodium chloride , per cent. Megatherium lysin , H.Is .
No. ... 1 .
2 .
2a .
2b .
3 .
4 .
5 .
1e- 7 .
8 .
; P. c ... .
0 *88 .
0*81 .
HBu .
15 .
I 11 .
1 a j 7-5 .
S. 23 -9 47 *1 101 *5 97-4 16 -7 0-2 0-7 0-5 o-o 0-5 o-o 31 *4 55 -1 103-3 97-8 41 -7 5-8 0-5 o-o o-o 5-0 3-5 39*0 76 1 99-6 96-0 47-8 60-4 0 0 2-3 0 0 17 -7 10-7 1 49-0 89*5 101 -1 93 1 56 -9 74-0 0 0 5-1 3-7 25 -0 | 17 -8 56*0 95 *3 99-6 94-2 65-6 85 -4 o-o 5-1 27-8 34-6 21-4 *\#169 ; 64-1 96 -7 97 -6 93 -5 70-6 88-5 0 0 7-1 46*3 45 -1 25-0 GO o 71 *1 97*6 100-7 95-7 76 -8 88-5 0 0 8-6 33 -9 48-8 I 28-5 76 -1 97*6 97-6 94-2 80-4 90-6 0 0 8-6 47-8 48-8 38 *5 \#171 ; A 79 -2 97*6 \#151 ; 94-2 83 -1 90-6 o-o 8-6 47-0 48-8 !
32-1 | 81 *7 99*2 \#151 ; 96 -0 85-0 90*6 o-o 12 -3 47-0 48-4 32-1 GO g 8 Xfl 83 -3 99-6 \#151 ; 94-9 86 -3 90-6 o-o 12 -8 45 -6 48-4 35-7 112 *6 100-0 96 -8 \#151 ; \#151 ; 2 -3 8-0 47-0 49-3 35-7 107*8 99 -6 96-4 \#151 ; \#151 ; 3-7 10 0 46-3 48-8 !
50-0 107 *8 99*2 \#151 ; 96'4 \#151 ; \#151 ; 10-0 \#151 ; \#151 ; 48-4 j 50-0 98*4 \#151 ; \#151 ; 96-4 \#151 ; \#151 ; \#151 ; \#151 ; \#151 ; \#151 ; !
50-0 96*2 93*7 50-0 Res. \lt ; 134 *9 100 *0 100 0 ioo-o 110 -o 125 -0 85-0 62-9 55-6 1 52 -7 67 -8 c.c. c.c. c.c. c.c. c.c. c.c. c.c. c.c. c.c. c.c. l 20 44 65 40 45 \#151 ; 47 45 34 46 31 f 100 *0 100 -o 100 -o 100-0 100 *0 100 -o 50 -0 50 -0 50 -0 50 0 50 -0 Ong.\lt ; c.c. c.c. c.c. c.c. c.c. c.c. c.c. c.c. c.c. c.c. l 100 112 112 120 105 \#151 ; 110 105 100 110 105 overnight , was poured out , and the filter case refilled with 112 c.c. of saline , of which 40 c.c. in 5 c.c. fractions were pressed through .
Examination for Sodium Chloride.\#151 ; Experiment No. 2a , Table I. Original Fluid : 0'81 per cent. NaCl = Orig .
( 100 ) .
Filtrates : The 1st and 2nd fractions contained ( 101*5 ) and ( 103*3 ) respectively , and those succeeding showed values varying about ( 100 ) .
1905 .
] Crystalloids and Colloids through Gelatine .
Eesidual Fluid : Titrated immediately after decompression ( 100 ) , after 12 hours in the filter case ( 110 ) .
Conclusion.\#151 ; On renewing filtration the filtrate may contain a higher percentage of salt than the original fluid , due probably to the temporary decompression , for on continued recompression the concentration tends to become less than ( 100 ) .
Continued Filtration of Sodium Chloride with Megatherium Lysin and Antilysin through 9 per cent. Gelatine .
In a former paper , * mixtures of lysin and antilysin were pressed through gelatine which had been tested for tightness by the filtration of saline .
The redistribution of salt under these circumstances was now determined .
Method.\#151 ; A filter through which 10 c.c. of 0'81 per cent. NaCl had been pressed , the original gelatine being 9 per cent , in 0'81 per cent NaCl , was decompressed , the residual salt solution removed , and the filter case refilled with 120 c.c. of a nearly neutral mixture of equal volumes of a fluid megatherium lysin and 5 per cent , antilysin in saline , of which 75 cc .
in 5 c.c. fractions were pressed through .
Examination for Sodium Chloride.\#151 ; Experiment No. 2b , Table I , is one of two practically identical filtrations .
Original Fluid : Contained 0'81 per cent. NaCl = ( 100 ) .
Haemolytic Index ( 16'8 ) .
Filtrates : As in No. 2a the 1st and 2nd fractions were the most concentrated in NaCl , but they did not reach ( 100 ) , and the final fraction was ( 96'4 ) .
The average haemolytic index of the whole filtrate was ( 0-5 ) .
Residual Fluid : The results were entirely similar to those of No. 2a as regards NaCl .
Haemolytic Index ( 38'6 ) .
Conclusion.\#151 ; The redistribution of salt is insufficient to give a trace of haemolysis under standard conditions .
Temporary decompression even for a few minutes allows a higher percentage of salt to pass into the filtrate than would be obtained by constant pressure .
Filtration of Sodium Chloride with Butyric Acid through 9 per cent. Gelatine .
Butyric acid in saline acts as a very strong haemolysin , and , as it possesses crystalloidal properties , it appeared to be of interest to compare its filtration phenomena with those of megatherium lysin in saline .
Further , butyric * Loc .
cit. Mr. J. A. Craw .
On the Filtration of [ Dec. 1 , acid greatly diminishes the surface tension of saline against air , and the possibility of a change in the surface forces between saline and gelatine formed a second point of interest , which is discussed towards the end of this paper .
Method.\#151 ; A solution of 0'8l per cent. NaCl containing 4'8 per cent , butyric acid was pressed through 9 per cent , gelatine in 0'81 per cent. NaCl at 100 atmospheres .
The content of the fractions of filtrate , etc. , in NaCl was determined as before , and the butyric acid was estimated by titration with standard sodium hydrate , using phenol-phthalein as indicator .
The haemolytic powers of the various fluids were also determined by the time taken for 1 c.c. of the fractions to hsemolyse completely 2 c.c. of a 2-5 per cent , suspension of guinea-pig 's corpuscles in 0-8 per cent. NaCl at 16 ' C. Examination for Sodium Chloride.\#151 ; Experiment No. 3 , Table I. The redistribution of salt was qualitatively similar to that which had been obtained in the absence of butyric acid .
The amount passing into the filtrate was , however , smaller .
Examination for Butyric Acid and Haemolytic Power.\#151 ; Original : 4'8 per cent , butyric acid = ( 100 ) HBu .
Haemolytic time = 8 minutes .
Filtrates : The 1st and 2nd fractions contained ( 0'2 ) and ( 5'8 ) HBu respectively .
The 1st fraction agglutinated the test blood strongly , and the 2nd haemolysed completely in 50 minutes .
The 8th to the 11th fractions contained ( 90-6 ) HBu , and the haemolytic time was 12 minutes .
Besidual Fluid : Concentration ( 125 ) HBu .
Haemolytic time , 5 minutes .
Conclusions.\#151 ; Butyric acid is retained by gelatine to a considerable extent .
The gelatine appears to retain more sodium chloride in the presence of butyric acid .
Filtration of Sodium Chloride through 9 per cent. Formalised Gelatine .
It seemed probable , from the marked change which takes place in gelatine on being exposed to formic aldehyde , that such gelatine would show a different permeability to that already found for ordinary gelatine .
On the other hand , the possibility of investigating physiological fluids in an apparatus which could be thoroughly and easily sterilised by means of " formalin " might recommend the use of formalised gelatine filters .
Method.\#151 ; A filter containing 9 per cent , gelatine in 0'8 per cent. NaCl was fitted into the filter case , and the latter filled with a solution of 08 per cent NaCl containing 10 per cent , of commercial " Formalin .
" A few cubic centimetres of fluid were pressed through , and the remainder allowed to stand in the filter case overnight .
The solution was then removed , and , after about 12 hours , 100 c.c. of 0'88 per cent. NaCl were placed in the filter case , and pressed through in fractions of 4 to 5 c.c. 1905 .
] Crystalloids and Colloids through Gelatine .
319 Examination for Sodium Chloride.\#151 ; Experiment No. 1 , Table I. Original Fluid : 0'88 per cent. NaCl represented as Orig .
( 100 ) .
Filtrates : No formaldehyde or formic acid could be detected .
From the 1st to the 11th fraction the Nad gradually increased from ( 23-9 ) to ( 83'3 ) , at 100 atmospheres filtration pressure .
At this point the pressure was removed and filtration resumed after an interval of eight hours .
Six fractions of 5 c.c. each were then removed , and gave values decreasing from ( 112-6 ) to ( 93*7 ) .
Eesidual Fluid : Tested immediately after the last decompression contained ( 134'9 ) .
Conclusions.\#151 ; Formalised gelatine containing sodium chloride retains sodium chloride from a solution , on filtration , to a greater extent than ordinary gelatine .
The decompression in this case also leads to the immediately following filtrates having a higher concentration of salt than the original fluid , but the effect is more marked than with ordinary gelatine .
The concentration of the residual fluid is also much greater than with ordinary gelatine .
Filtration of Megatherium Lysin Through Various Percentages of Gelatine .
Megatherium lysin diffuses slowly compared with crystalloids in general , * and is probably of semi-colloidal character .
It seemed probable that a substance of this type would be considerably more affected by a change in the concentration of gelatine than a good filter-passer such as sodium chloride .
' Hardy 's ( 1899 ) work on gelatine pointed also to the possibility of considerable change in the structure of the jelly at about 7 per cent , gelatine , which might lead to markedly different degrees of permeability between 7 and 15 per cent gelatine filters .
Method.\#151 ; Filters containing 15 , 11 , 9 and 7*5 per cent , gelatine in 0-8 per cent. NaCl were tested for tightness with 08 per cent. NaCl , of which 10 c.c. were pressed through in each case to ensure a concentration of NaCl in the succeeding filtrates of about 0'8 per cent. The fluid examined was the haemolytic filtrate from a broth culture of B. megatherium diluted with an equal volume of 0'8 per cent. NaCl .
Filtration was carried through at a pressure of 100 atmospheres and a temperature of 10 ' to 12 ' C. and the filtrate was collected in fractions of about 4 c.c. Examination for Lysin.\#151 ; Table I. Original Fluid.\#151 ; The haemolytic index ( see above ) was ( 50 ) .
* Cf .
Craw ( IY , 1905 ) .
Mr. J. A. Craw .
On the Filtration of [ Dec. 1 , Filtrates : Experiments Nos. 4 , 5 , 6 and 7 , Table I , show the haemolytic values of the succeeding fractions through 15 , 11 , 9 and 7'5 per cent , gelatine respectively .
The average indices for the whole filtrates were in the order given above ( 1*5 ) , ( 6-9 ) , ( 31*7 ) , and ( 36'9 ) .
In Experiment No. 8 , Table I , 50 c.c. of 20 per cent , horse serum in 08 per cent. NaCl had previously been passed through a 7'5 per cent , gelatine filter .
In this case the permeability to lysin was slightly less than that of a similar filter without serum , the average indices of the filtrates being ( 306 ) and ( 36-9 ) respectively .
Gelatine : On melting out the gelatine , after filtration , at 37 ' C. , and mixing with the test blood , the latter was in all cases rapidly and completely hsemolysed , index ( 100 ) .
Control experiments showed that saline on being pressed through gelatine which had been used to filter lysin became strongly haemolytic .
The original gelatine had no haemolytic effect in the standard time .
Residual Fluids : The residual fluids in all cases showed average indices which were higher than that of the original fluid .
The lowest portions of residual fluid had generally higher indices than the portions towards the upper surface .
With rising concentration of gelatine the haemolytic power increased , thus with 7'5 per cent , the residue had an index ( 52-7 ) whereas with 15 per cent , it was ( 85 ) .
Conclusions.\#151 ; Megatherium lysin is retained to a greater extent than sodium chloride , and more is retained with higher concentrations of gelatine .
The residual fluids have higher concentrations than the original and the concentration is greater with higher percentages of gelatine .
Diffusion and expression of lysin from the gelatine into the residual fluid are insufficient to account for the increased concentration immediately after decompression , and it seems as if the water of the original fluid could pass into the gelatine more readily than the lysin .
It seems possible , under the conditions of preparation , that during the draining of the filters from the various percentages of gelatine at 30 ' C. , the filters with the higher concentrations would retain thicker surface layers of gelatine and so exaggerate the differences in permeability .
Filtration of a Typical Colloid , Ferric Hydrate , through 11 per cent. Gelatine under Varying Pressure .
As ferric hydrate in colloidal solution shows no appreciable tendency to diffuse , any redistribution of the colloid by the gelatine filter must be explained on some other basis than that of diffusion .
Method.\#151 ; A 5-per-cent , solution of colloidal ferric hydrate was prepared 1905 .
] Crystalloids and Colloids through Gelatine .
321 by adding ammonium carbonate to ferric chloride and dialysing for three weeks ; it gave no trace of red coloration with potassium thiocyanate , but showed the characteristic reaction of colloidal ferric hydrate , viz. , a slight yellowish precipitate .
The filter used contained 11 per cent , gelatine which had been dialysed for 24 hours .
The content of the fractions of filtrate , etc. , , in iron , was determined by converting the hydrate into chloride and estimating colorimetrically with potassium thiocyanate .
Table II .
Colloidal ferric hydrate .
Neutral red .
Iodine in potassium iodide .
Yol .
in c.c. Pressure in atmos .
Per- centage Fe .
Vol. in c.c. Pressure in atmos .
Per- centage N.R. Yol .
in c.c. Pressure in atmos .
Per- centage i* .
Per- centage KI .
4 100 o-o 4 100 o-o 1 100 5 68 -8 i j 4 100 0-7 4 100 2*3 4 100 8 69 -9 4 100 0-6 4 100 0*04 4 100 17 70-9 \#163 ; 4 100 0-4 3 20 0*03 4 100 19 69-9 : 4 100 0-4 10 100 0*30 4 100 21 69-9 *8 S 5 4 100 0-4 10 100 0*78 4 100 23 69-9 4 100 o-i 10 100 0*90 4 100 25 68-3 4 100 o-o 10 100 1*17 1 15 20 69-9 ; 4 100 o-o 10 100 3*00 5 100 33 71*5 | 1 1 100 o-o 10 100 3*20 5 100 33 73-0 1 1 !
m 0-5 20 250 -0 20 100 4*68 4 100 35 72-0 i 1 100 4-75 \#151 ; \#151 ; \#151 ; 0*75 0 0 43-0 2 100 0-9 \#151 ; \#151 ; \#151 ; 0*75 100 25 93 -5 3 50 0-7 \#151 ; \#151 ; \#151 ; 6 100 37 90-5 0-5 15 66 -6 \#151 ; \#151 ; \#151 ; 6 100 50 86 -0 0-75 50 1 -o \#151 ; \#151 ; \#151 ; 1 0 1 34 -4 0-25 30 0-6 \#151 ; \#151 ; \#151 ; 5 10 25 94-6 0-5 20 0 3 \#151 ; \#151 ; - \#151 ; t 2 100 30 92 -5 0 75 10 0-1 \#151 ; \#151 ; \#151 ; 3 50 30 96-8 : 0-25 50 0-2 \#151 ; \#151 ; 4 100 30 91 -4 1 100 0 -2 _ \#151 ; 5 50 30 85 -0 5 15 0-4 \#151 ; \#151 ; \#151 ; 5 25 30 82 -8 4 100 0 '4 \#151 ; \#151 ; \#151 ; 5 25 25 79 -5 0-5 i 4 15 100 0-4 0-4 5 25 25 73 -1 Res. 50 \#151 ; 180 -0 20 \#151 ; 18 -75 35 I ' \#151 ; 30 109 *7 Orig .
| 120 \#151 ; 100 -o j 120 100 -o 120 \#151 ; 100 100 *0 Examination for Ferric Hydrate.\#151 ; Table II gives the relations obtained in one out of three similar experiments .
Original Fluid : Five per cent , ferric hydrate = ( 100 ) Fg .
Filtrates : The 1st and 2nd fractions contained ( 0-0 ) and ( 0'7 ) respectively , and in the succeeding fractions the amount diminished until in the 8th no Mr. J. A. Craw .
On Filtration of [ Dec. 1 , trace could be found at 100 atmospheres filtration pressure .
On suddenly-diminishing the pressure to 20 atmospheres the liquid percolating through was very intensely coloured and gave the value ( 250 ) .
Ee-establishing the pressure of 100 atmospheres , the filtrates became less and less intensely coloured ( 4*75 ) and ( 0'9 ) .
A drop in the pressure to 50 atmospheres did not materially change the concentration , but a further drop to 15 atmospheres gave a filtrate with the value ( 66-6 ) .
Increasing the pressure to 50 atmospheres caused the filter to tighten once more with respect to the hydrate , and on gradually decreasing the pressure to 10 atmospheres the filter remained tight .
In this condition sudden variations of pressure from 100 to 15 atmospheres had but a slight effect on the permeability of the filter .
Eesidual Fluid : The colour was more intense than that of the original fluid and its content was ( 180 ) Fe .
Conclusions.\#151 ; The gelatine After is slightly permeable to the typical colloid ferric hydrate , but at constant pressure the permeability decreases as the filtration proceeds .
The permeability is increased enormously by suddenly diminishing the pressure , but is not much affected by a gradual diminution .
After gradual diminution of pressure a filter is obtained which does not markedly change in permeability on suddenly varying the pressure .
The high concentration of the residual fluid is probably due to the water penetrating the gelatine easily , whereas the ferric hydrate is largely left on the surface of the gelatine , where it forms a concentrated solution of higher specific gravity than the rest of the fluid and so gives rise to convection currents which cause it partly to mix with the remainder of the fluid in the filter case .
The candle retains a skin of colloidal ferric hydrate , and the filter most probably at the beginning of the filtration acted as a simple gelatine filter , but subsequently as a compound ferric hydrate gelatine filter .
Filtration of a Staining Colloid , Neutral Fed , through 11 per cent. Gelatine under Varying Pressure .
* Preliminary experiments with horse serum and soluble starch showed qualitatively similar effects to those obtained with colloidal ferric hydrate , and it seemed probable that the majority of colloidal solutions would behave in a similar way on filtration through gelatine .
It seemed probable , however , that those colloidal solutions which stain gelatine would show considerable difference in behaviour .
Method.\#151 ; A 0'5-per-cent , solution of neutral red in distilled water was pressed through 11 per cent , dialysed gelatine and the content of the fractions of filtrate , etc. , determined colorimetrically .
1905 .
] Crystalloids and Colloids through Gelatine .
323 Examination for Neutral Bed.\#151 ; Table II .
Original Fluid : 05 per cent , neutral red = ( 100 ) N.R. Filtrates : The 1st , 2nd and 3rd fractions showed ( 0-0 ) , ( 2%3 ) and ( 0'04 ) respectively .
A sudden diminution in pressure from 100 to 20 atmospheres caused a slight decrease in the value of the filtrate ( 0'03 ) .
On raising the pressure once more to 100 atmospheres , the succeeding fractions gradually increased in content up to ( 4*68 ) N.R. Residual Fluid : The colour intensity was much diminished and indicated only ( 18'75 ) N.R. Conclusions.\#151 ; The gelatine filter is slightly permeable at the beginning of filtration to neutral red , but the permeability decreases , as in the case of colloidal ferric hydrate , on continued filtration at constant pressure .
Diminution of pressure has the opposite effect to that obtained with sodium chloride and ferric hydrate as the permeability tends to decrease .
The filter gradually becomes stained throughout and the permeability increases correspondingly .
Filtration of a Staining Crystalloid , Iodine in Potassium Iodide , through .
11 per cent. Gelatine , under Varying Pressure .
Iodine was chosen as a crystalloidal substance having the property of staining gelatine , and potassium iodide was used as its solvent in water .
The interest of the experiment centred chiefly in the relative behaviour of the two substances and the mutual influence they exert on each other during filtration under varying pressure .
Method.\#151 ; A solution of 0-214 per cent , iodine in 1'54 per cent , potassium iodide was filtered through 11 per cent , gelatine in 1*54 per cent , potassium iodide .
The content of the fractions of filtrate , etc. , was determined by titration with sodium thiosulphate , using starch as an indicator for the iodine and with silver nitrate , using potassium chromate as indicator for the potassium iodide .
Examination for Iodine and Potassium Iodide.\#151 ; Table II .
Original Fluid : 0214 per cent. I2 = ( 100 ) I2 .
1'54 per cent. KI = ( 100 ) KI .
Filtrates : The 1st c.c. contained a considerable percentage of iodine ( 5 ) and likewise of potassium iodide ( 68'8 ) .
On continued filtration at constant pressure the iodine concentration rose steadily to ( 25 ) ; the iodide rose slightly and then fell to ( 68'3 ) .
On suddenly , diminishing the pressure from 100 to 15 atmospheres the iodine concentration diminished and the iodide increased .
On raising to 100 atmospheres the iodine rose , whereas the iodide rose and then fell in concentration .
Removing the pressure for 12 hours the few drops of liquid which percolated through contained no iodine and only ( 43 ) of 324 Mr. J. A. Craw .
On the Filtration oj [ Dec. 1 , iodide .
Re-establishing the pressure of 100 atmospheres the iodine content rapidly increased , whereas the iodide in the first few drops was abnormally high ( OS'S ) , but decreased in the second fraction of 6 c.c. to ( 86-0 ) .
Once more removing the pressure for three hours the iodine diminished to ( 1*0 ) and the iodide to ( 34'4 ) , confirming the former result .
The pressure was now slightly raised , viz. , to 10 atmospheres , when the iodine became ( 25 ) and the iodide reached the highest value so far ( 94'6 ) .
A further rise in pressure to 100 atmospheres increased the iodine to ( 30 ) and diminished the iodide to ( 92*5 ) .
Variation between 100 and 50 atmospheres did not further affect the iodine content but did influence the iodide .
Finally , a gradual fall in pressure to 25 atmospheres caused both iodine and iodide to diminish in concentration .
Residual Fluid : The concentration of iodine was only ( 30 ) , whereas the iodide was ( 109'7 ) .
Conclusions.\#151 ; The gelatine absorbs a large amount of iodine , but is easily permeable to the same and the permeability rises and falls , as in the case of neutral red , with rising and falling pressure .
The increasing concentration of iodine in the gelatine increases the power of the latter in retaining potassium iodide .
On entirely removing the applied pressure the gelatine robs the liquid percolating through almost entirely of its iodine and the iodised gelatine abstracts about two-thirds of the iodide .
The filtrates , after a drop in the pressure , show an increased amount of iodide on recompression similar to the effect observed in the filtration of sodium chloride .
On the Influence of the Nature of the Solution and of Varying Pressure on the Rate of Filtration .
A few preliminary measurements of the rates of filtration of various solutions were made to elucidate some of the physical properties governing the action of the gelatine during filtration .
Method.\#151 ; As a rough approximation the number of drops per minute falling from the nozzle of the filter case was taken as a measure of the rate of filtration .
The rate was determined after the first 2 c.c. had passed , and again after about 50 c.c. in all had been collected .
Rates of Filtration.\#151 ; Table III shows that distilled water , 0'8 per cent , sodium chloride and megatherium lysin pass 9 per cent , gelatine in the initial stage more rapidly than in the final under constant pressure .
Solutions containing 0'214 per cent , iodine , with 1'54 per cent , potassium iodide , and 5 per cent , colloidal ferric hydrate pass 11 per cent , gelatine with similar variations in rate .
For all these solutions the decrement in rate is practically 1905 .
] Crystalloids and Colloids through Gelatine .
325 the same .
On filtering horse serum in full strength , however , the decrement in rate was found to be abnormally large .
Table III .
100 atmos .
12 ' C. NaCl .
Kb I. NaCl lysin .
NaCl toluol .
NaCl Am .
Ale .
NaCl HJBii .
Dist .
water .
Coll. Fe(HO)3 .
Serum .
Initial drops per minute 7 7 8 16 18 6 9 6 6 Final drops per minute \#187 ; 4 4 5 18 60 18 4 4 1 A solution of 0'8 per cent , saline containing toluol as a fine emulsion did not show any decrement .
On the other hand , 08 per cent , saline containing either amylic alcohol or butyric acid ( 4'8 per cent. ) gave very distinct increments in the rate of filtration .
Table IY gives a comparison of the rates obtained for distilled water and 081 per cent , sodium chloride containing 4'8 per cent , of butyric acid .
It is evident that distilled water filters more and more slowly , whereas the saline butyric acid passes more and more rapidly .
The effect of removing the pressure for two minutes enables the distilled water on re-establishing the pressure to pass through at a higher rate , and if the pressure be removed for 10 minutes the gelatine regains completely its original rate of filtration .
Table IV .
Distilled water .
Sodium chloride and butyric acid .
Pressure Time in Drops per C.c. per Pressure Time in Drops per C.c. per in atmos .
minutes .
minute .
minute .
in atmos .
minutes .
minute .
minute .
I 100 0 7 100 0 6 100 13 6 0*5 100 2 5 \#151 ; 100 21 5 0-5 100 5 9-5 \#151 ; 0 23 0 \#151 ; 100 7 11 0*7 100 29 6-5 0*6 95 15 9-75 0-6 100 32 5 0*5 90 20 9 \#151 ; 100 40 4-5 0*3 100 23 11 0-6 0 50 0 \#151 ; 100 35 13 0-6 100 73 9 0-7 100 42 14 0-7 100 83 5 -5 0*5 100 49 15 0-7 100 133 4 5 0*4 100 64 16 0*8 100 71 18 0-9 100 74 18 1 1 Mr. J. A. Craw .
O the Filtration of [ Dec. 1 , Conclusion.\#151 ; Solutions of salts such as sodium chloride , or colloids such as ferric hydrate , will not appreciably modify the rate of filtration through gelatine unless in concentrated solution .
Amylic alcohol and butyric acid accelerate the rate of filtration .
It seemed possible that this effect might be due to the condensation of amylic alcohol and butyric acid on the walls of the gelatine pores , and that the change in viscosity might account for the increased rate .
This , however , does not seem probable , for a toluene emulsion did not markedly affect the rate of filtration , and as the viscosities of the toluene and amylic alcohol used were found to be 0*49 and 4-4 respectively when compared with distilled water as unity , one might conclude that the toluene would accelerate , and the amylic alcohol retard , the rate of filtration .
On the other hand , the action of amylic alcohol and butyric acid is consonant with the view that the surface forces between gelatine and saline are modified .
The surface tension of saline against air diminishes with increasing concentration of both amylic alcohol and butyric acid .
Both of these substances markedly influence the rates of sedimentation of solid suspensions through water , * and this fact seems to have a close connection with their influence on the rate at which water passes through a porous solid or jelly such as gelatine .
The effect of continued pressure on gelatine seems to be a tightening of the pores , which is nearly complete at 100 atmospheres in 40 minutes , and decompression apparently allows the pores to resume their original dimensions in about 10 minutes at atmospheric pressure .
On the Expression of Water and Gelatine from a Wet Filter .
In agreement with E. W. Reid , the filtrates were found to contain gelatine , but in variable amounts .
Thus the first fraction of about 4 c.c. usually contained most , and after 20 c.c. had passed the gelatine was only present in traces .
Rough estimations with tannic acid indicated about 0'5 per cent , gelatine in the first , and less than 0T per cent , in the fifth fraction on filtering distilled water .
These contents of gelatine were not found to influence the titrations given above .
It has been hitherto assumed that the water in the pores of a wet filter is largely expressed before the liquid undergoing filtration appears in the filtrate undiluted by the same .
For this reason Martinj* discarded the first 10 c.c. of filtrate , and E. W. ReidJ concluded that the concentration of even a second fraction of 25 c.c. might be considerably reduced by this dilution .
It seems to me , however , that the effect of the water in the gelatine on the concen-* Cf .
Craw ( 1904 ) .
t cit. J Loc .
cit. 1905 .
] Crystalloids and Colloids through Gelatine .
327 tration of the filtrate is neither so marked nor so long continued as has been imagined , for the following reasons .
In the first place , the filtrates from saline through gelatine containing the same concentration of saline are at first only about half as concentrated as the original saline .
Secondly , on decompression , and again filtering , the filtrate may contain a higher concentration of salt than the original saline .
This seems conclusively to show that much of the dilution of the filtrate is due not to the water of the filter , but to the retention of the salt by the gelatine .
Further , Reid found the quantity of water in freshly prepared filters to vary between 2 and 13 grammes\#151 ; presumably , the amount which can be removed by dry air .
It is probable that only a small part of this water will be removed by the passage of a slow current of liquid through the gelatine , the remainder being retained by adsorption or inhibition forces , etc. In this connection an observation which was made with all the filtrations given in this paper may not be unimportant .
It was found that the sum of the volumes of the filtrate fractions and residual fluid in any one filtration experiment was less than the volume of the original fluid introduced into the wetted apparatus .
It is a well-known fact that water can be forced into gelatine by pressure , and as a large part of the gelatine in the pores of the filter , during filtration at 100 atmospheres , must be under considerable pressure , it seems reasonable to assume that part of the original water of the wet filter is more firmly bound , and that about 5 c.c. are imbibed partly from the water of the filter and partly from the fluid filtered .
The rapid passage of iodine , neutral red , and colloidal ferric hydrate into the filtrate also point to a considerable percentage of the first fractions of filtrate being contributed by the fluid filtered .
It seems , on the whole , as if the free water of the filter is almost wholly removed in the first fraction of 5 c.c. filtrate .
The Application of Mechanical and Adsorption Hypotheses to the Filtration Phenomena of Gelatine .
Mechanical Hypotheses.\#151 ; The most obvious explanation of the retention of colloidal substances by the gelatine filter is that the colloidal molecule or grain is too large to pass through the pores .
Martin* has advanced the view that the non-filtration may be due either to the size of the molecules or to some interaction between the colloidal membranes and the dissolved colloidal molecules .
It has , however , been shown by E. W. Reid ( 1904 ) , Gatin-Grazewska ( 1904 ) , and others that proteids , glycogen , and other typically colloidal substances , exert no measurable osmotic pressure in solution , do not influence the freezing point of the aqueous medium , and in general diffuse Mr. J. A. Craw .
On Filtration of [ Dec. 1 , very slowly , if at all .
These facts lend much probability to Oswald 's view that colloidal solutions may be regarded as very fine suspensions .
As the molecular weights of these substances are as yet quite indefinite , it would seem more justifiable to speak of the colloidal " grain " than of the colloidal " molecule .
" Accepting this view of the nature of colloidal solutions , the action of the filter admits of an apparently satisfactory explanation on a purely mechanical basis.* During the filtration , for example , of colloidal ferric hydrate , Table II , the first fraction contained none of the colloid , and probably consisted chiefly of the water of the filter .
In the succeeding fractions a little ferric hydrate passed through the larger pores , but these were speedily blocked by the particles .
Under the pressure of 100 atmospheres the pores are probably compressed , and the effective passages are soon completely blocked by ferric hydrate .
On suddenly diminishing the pressure the gelatine begins to return to its original condition , and the effective passages increase in number and diameter .
The condensed ferric hydrate , no longer completely blocking the widened passages , is swept out of the gelatine by the oncoming fluid , and so gives a filtrate with an abnormally high concentration .
On the other hand , by gradually diminishing the pressure the ferric hydrate has time to fill the new and the gradually-widening pores , so that the filtrate contains only a trace of colloid .
If the ferric hydrate be not expressed on recompression , it is conceivable that the compound filter of ferric hydrate and gelatine would be less affected as regards permeability by variation of pressure than the original gelatine .
Much as this view has to recommend it , it does not seem to be the whole explanation , as it does not satisfactorily account for the phenomena observed with sodium chloride .
It is highly improbable that sodium chloride is retained by a mechanical blocking of the passages , and that it is swept in high concentration into the filtrate , on decompression , because the condensed sodium chloride is no longer of the same dimensions as the widened passages .
Eecourse must , therefore , be had to some other view , which may considerably modify the explanation given for colloidal substances .
Adsorption Hypotheses.\#151 ; Porous substances , powders , and fine suspensions of solids in aqueous media have the power of removing salts and other substances from solution .
The action is attributed to forces of the same nature as those which give rise to adhesion and the wetting of a solid by a liquid , e.g. , of clean glass by water .
The phenomenon usually called adsorption is common to all substances with highly developed surfaces , and * Cf .
also M. Traube ( 1866 , 1867 ) .
Traube regards precipitation membranes as atomic and molecular sieves .
Crystalloids and Colloids through Gelatine .
the intensity of the adsorption depends not only on the physical condition but also on the chemical nature of the active surface .
Gelatine being , probably , a porous substance , presents a large surface to the enclosed fluid , and it seems highly probable that when this fluid contains crystalloids or colloids adsorption will take place .
Thus , gelatine containing 08 per cent , saline apparently retains about one-half of its content in sodium chloride , and the retention of crystalloids and colloids on filtration through gelatine at constant pressure seems to admit of explanation on the basis of adsorption .
The results obtained in the investigation of the adsorption phenomena of the sedimentation of silts through aqueous solutions may therefore be applied to what apparently is the converse of that process , viz. , the passage of aqueous solutions through porous solids or powders , under pressure .
This is rendered highly probable from the fact that the rate of sedimentation of a silt through saline is generally accelerated by the presence of butyric acid , amylic alcohol , and other substances influencing surface tension , and similarly the rate of filtration through the porous gelatine is accelerated by the same substances .
An additional factor may , however , come into play in gelatine filtration .
The effect of pressure on gelatine containing a solution of salt will be to cause the absorption of water .
It seems permissible to assume , in view of Hardy 's work on the structure of gelatine , that this additional water will be taken up by the web mass and will probably lead to a diminution in dimensions of the pores or web spaces .
This view would account for the continued diminution in rate of filtration at constant pressure .
The process should show some similarity to the passage of a solution into filter paper , where the water passes in more readily than the substance in solution .
A sudden diminution in pressure will lead to the rapid expulsion of the imbibed fluid , which will sweep the adsorbed matter into the widening pores .
The gelatine itself will thus express part of this concentrated fluid , and at low pressures highly concentrated filtrates will be obtained .
This will take place markedly when the adsorption is easily reversible , but the less reversible adsorptions of certain staining substances , neutral red and iodine will not give this effect .
Further compression and decompression would presumably lead to a more uniform distribution of the adsorbed substance throughout the gelatine , and as the filter is now also more saturated , the effect of variation of pressure on the concentration of the filtrate would be less marked .
Summary of Conclusions .
1 .
Wet gelatine filters are to be preferred to those which have been partially dried , as the former have more uniform rates of filtration , and dilution of the filtrate by the water of the gelatine can be largely eliminated .
Mr. J. A. Craw .
On , the Filtration of [ Dec. 1 , 2 .
Under constant pressure the gelatine of the filter absorbs water , and its porosity gradually decreases ; on decompression this water is expressed , and the original porosity is rapidly regained .
3 .
Gelatine of a definite concentration apparently has a specific permeability for different crystalloids and colloids ; the value is high but not complete for the crystalloids sodium chloride , potassium iodide , and butyric acid , and it is low , but not zero , for the colloids ferric hydrate , serum and soluble starch .
4 .
As filtration proceeds the crystalloids pass through in increasing concentration , whereas the colloids rapidly decrease to zero .
5 .
The simultaneous filtration of two substances may influence their specific permeabilities , thus butyric acid alters the permeability to sodium chloride , and iodine that of potassium iodide .
6 .
Variations in the gelatine influence the permeability , e.g. , formalised gelatine is less permeable to sodium chloride than ordinary gelatine , and 15 per cent , gelatine is less permeable to megatherium lysin than 7'5 per cent. 7 .
Variation in the pressure causes remarkable changes in permeability .
A sudden diminution of pressure gives highly concentrated filtrates of both crystalloids and colloids , whereas a gradual diminution has practically no effect .
8 .
Substances which stain the gelatine , e.g. , neutral red and iodine , give filtrates with lower concentrations on diminishing the pressure .
9 .
The rate of filtration is accelerated by amylic alcohol and butyric acid , which accelerate the rate of sedimentation of silts in a similar way .
10 .
Part of the phenomena may be explicable on the mechanical view of a blocking of the gelatine pores , but chemical relations between the gelatine and substances filtered must be taken into consideration , and probably the most satisfactory view is that the action of gelatine on the solutions tested is essentially an adsorption phenomenon .
Crystalloids and Colloids through Gelatine .
Arrhenius and Madsen ( 1902 ) , ' Festskrift ved Indvielsen af Statens Seruminstitut , ' III .
Arrhenius and Madsen ( 1904 ) , ' Bulletin de l'Academie Royale des Sciences de Dane-mark , ' p. 271 .
Bordet ( 1903 ) , 'Annales de l'Institut Pasteur , ' vol. 17 , p. 161 .
Brodie ( 1897 ) , ' Journ. of Pathology , ' p. 460 .
Brodie ( 1900 ) , ' Brit. Med. Journ. , ' p. 300 .
Craw ( 1904 ) , ' Lancet , ' p. 434 .
Craw ( I , 1905 ) , ' Journ. of Hygiene , ' vol. 5 , p. 115 .
Craw ( IY , 1905 ) , 'Roy .
Soc. Proc. , ' vol. 76 B , p. 179 .
Ehrlich ( 1898 ) , ' Deutsche Med. Wochenschrift , ' vol. 24 , p. 597 .
Ehrlich ( 1903 ) , ' Berliner Klin .
Wochenschrift , ' vol. 40 , pp. 793 , 825 , 848 .
Gatin-Grazewska ( 1904 ) , ' Pfliiger 's Archiv , ' vol. 103 , p. 281 .
Hardy ( 1899 ) , ' Journ. of Physiology , ' vol. 24 , p. 158 .
Kralft ( 1902 ) , ' Zeitschr .
f. physiolog .
Chemie , ' vol. 35 , pp. 364 , 376 .
Landsteiner ( 1903 ) , 'Miinchener Med. Wochenschrift , ' vol. 50 , p. 764 .
Martin , C. J. ( 1896 ) , 'Journ .
of Physiology , ' vol. 20 , p. 364 .
Martin and Cherry ( 1898 ) , ' Roy .
Soc. Proc. , ' vol. 63 , p. 420 .
Nernst ( 1904 ) , ' Zeitschrift fur Elektrochemie , ' vol. 10 , p. 377 .
Reid , E. W. ( 1901 ) , 'Journ .
of Physiology , ' vol. 27 , p. 161 .
Reid , E. W. ( 1904 ) , 'Journ .
of Physiology , ' vol. 31 , p. 438 .
Starling ( 1889 ) , ' Journ. of Physiology , ' vol. 24 , p. 317 .
Traube , M. ( 1866 ) , 'Centralbl .
f. d. Med. Wissenschaften , ' vol. 4 , pp. 91 , 113 .
Traube , M. ( 1867 ) , 'Archiv f. Anat .
u. Physiologie , ' p. 87 .

