Reprinted from the American Chemical Journal, Vol. XI, No. 6. 
DYNAMICAL THEORY OF ALBUMINOID AMMONIA.1 
Robert B.'Warder. 
The following investigation was suggested by committee work 
of the American Association for the Advancement of Science, on 
water analysis. It is an attempt to apply well established prin- 
ciples of mass action to certain questions involved in the determi- 
nation of albuminoid ammonia. The following points have been 
kept in view: 
1. Can some probable correction be made for the error incurred 
by stopping the distillation while ammonia is still coming off? 
2. What modification of the original method seems rational? 
3. What insight can be gained into the nature of the chemical 
process ? 
The experimental basis of the following paper will be found in 
Mallet's report on the determination of organic matter in potable 
water,2 with the addition of some details kindly communicated by 
Dr. Charles Smart. 
I. Distillation of Free Ammonia. 
In the distillation of albuminoid ammonia, two stages must be 
distinguished: first, oxidation of the nitrogenous body by the 
alkaline solution of permanganate; and, second, the separation of 
1 Prepared for the Toronto meeting of the American Association for the Advancement of 
Science. 
2 Report of the National Board of Health for 1882. 366 
Warder. 
ammonia thus formed by distillation. It will be most convenient 
to begin with the discussion of this second process, which is essen- 
tially physical. As in any other case of fractional distillation, 
water and ammonia both pass from the fluid to the gaseous or 
vapor condition, in accordance with their respective tensions and 
mass. 
Let x - volume of fluid in retort, 
y - weight of ammonia in retort, 
z - weight of ammonia in distillate. 
V 
Then - = concentration of fluid in retort, 
x ' 
- concentration of any small portion of dis- 
tillate. 
The coefficient of volatility may be defined as the ratio of the 
concentration in any small portion of distillate to that of the fluid 
in retort. Designating this coefficient as k, 
(0 
dx x v 
In the ordinary distillation of ammonia, k is found to be approxi- 
mately constant, and it will be so regarded here. By integration, 
loge_y = k. \oge x -f- constant; 
using zero subscript to denote initial conditions, 
logejKo = k. loge + constant, 
••• loge ~ = k . loge - J , 
_/o 
or log -=7?.log~, (2) 
Jro 
a very useful formula for determining the coefficient of volatility 
and tracing the progress of distillation. Curve A, in the figure, 
represents y as that function of x defined by equation (2). 
Wanklyn,1 who was the first to use and define the term coefficient 
of volatility, reports four experiments for its determination. One 
liter of dilute aqueous ammonia was distilled in each case until 
50 cc. of distillate had been collected. As nearly one-half the 
ammonia taken was distilled with the first 5 per cent, of water (the 
concentration in the retort being gradually reduced at the same 
1 Phil. Mag. [4], 45,132 (Feb. 1873). Dynamical Theory of Albuminoid Ammonia. 
367 
time), Wanklyn estimated k equal to " about thirteen or fourteen." 
The experimental data are given below with more precise values 
of k, as calculated by equation (2). In each case, x0= 1000, 
jv -950. The quantities of ammonia taken, and that found in 
the distillate, are expressed in milligrams under _y0 and z respec- 
tively ; the difference (remaining in retort) undery: 
Jo- Z. y. k. 
IOOO 480 520 12.74 
i -5° -50 13-50 
•5 -235 -265 12.39 
.2 .095 .105 12.55 
Mean - 12.8 
This value may be rather too low, as no account is taken of 
a possible loss from imperfect condensation, a loss averaging 7.2 
per cent, in Smart's experiments. 
The same author observes also1 that in the distillation of one- 
half liter of water, the ammonia found in the first measure of 50 cc. 
is three-fourths of the whole quantity of ammonia obtained. By 
equation (2), assuming that any loss affects each part in the same 
proportion, 
log 1 = k . log , 
••• £= 13-14- 
Thirty experiments by Smart to determine the loss of ammonia 
by imperfect condensation are also available for the estimation of 
k, the successive measures of distillate having been nesslerised 
separately.2 A single distillation affords several values, as each 
fraction determined may be compared with the whole quantity 
found in subsequent fractions. The values thus deduced vary 
from 10 to 16, with a few instances beyond these extremes. The 
variations in the values of k do not show any well marked relation 
to the whole quantity of ammonia present, to the amount of loss, 
nor to the rate of boiling. The experimental work was done 
wholly for another purpose, and probably with just such care as 
the conscientious analyst would ordinarily give. We may then 
assume 13 as a mean value for k, remembering that considerable 
allowance must be made for unexplained variations. 
J1 Wanklyn's Water Analysis, 6th and 7th editions, p. 41. 
2 Most of these data appear in the Report of the National Board of Health for 1882, pages 
2-3i4- 368 
Warder'. 
II. Application of Formula to Estimation of Free Ammonia. 
In the analytical process it is usual to distill three or four frac- 
tions, each nearly equal to one-tenth of the original volume. For 
100 parts of free ammonia present, when Z.'-13, theory would 
lead us to expect, in four successive fractions, the increments 
named below under with loss of 0.13; vertical lines in the 
figure, a, a, a, represent these increments geometrically. The 
ratio of each increment to the next is stated under r. The mean 
values of the ratios obtained in the experiments to determine loss 
of ammonia are given under r'\ 
No. 
I 
Az. 
74.6 
3-75 
4-23 
2 
19-9 
4-4 
5-26 
3 
4-53 
5-4 
5-65 
4 
.84 • 
In the analyses of natural waters reported foe. cit. pp. 321-330), 
considerable variation appears in the ratios of successive incre- 
ments, but usually not more than might be expected from the 
observed variations in k. Analyses No. 18 and No. 54 are excep- 
tional and happen to give identical results, the free ammonia 
being .015, .01, and .01 mgm. in the three fractions nesslerised, 
with ratios 1.5 and 1. In such cases there is a strong probability 
that some more complex body is gradually decomposed by simple 
boiling with sodium carbonate. Small ratios between the succes- 
sive fractions of " free ammonia " may lead to suspicion of urea 
contamination, though of course due caution should be used in 
drawing such conclusions. The analyst will at least find it instruc- 
tive to enter the ratios in his note-book, together with the sepa- 
rate results of nesslerising; this will appear more plainly in the 
following sections. The slowness of the chemical action usually 
retards the distillation of albuminoid ammonia, giving rise to much 
smaller ratios between successive fractions than those named above. 
III. Action of Mass in Formation of Albuminoid Ammonia. 
We must now attempt to follow the chemical as well as the 
physical aspect of the process. In the former the reactions may 
be much involved, and various formulas may be required to rep- 
resent the successive formation and destruction of different sub- 
stances. Let us consider, as an ideal case, a single chemical Dynamical Theory of Albuminoid Ammonia. 
369 
reaction, involving one molecule each of three reacting bodies; the 
formation of ammonia taking place at each moment of distillation, 
in each cubic centimeter of the fluid, in proportion to the product 
of the quantities of the nitrogenous body, the potash, and the 
permanganate present therein. 
Retaining the use of x,y, and z as explained in section I, let 
u - weight of the nitrogenous body in the retort, 
v ■=. weight of potash in the retort, 
w - weight of potassium permanganate in the retort, 
/ = time, in minutes, from the beginning of the reaction, 
a - coefficient of speed. 
Assume also a uniform rate of distillation, in which 
b~ volume distilled per minute, 
be - volume of fluid in retort at the beginning of the action; 
whence 
x-bc - bt, 
and dx - - bdt. 
V W 
Then - , --, and - = weights of the three active substances in 
x x x ° 
each cubic centimeter of fluid in the retort, 
- du , du c . 
- - b - rate of change, 
and, by the law of mass action, 
,du u v w u 
b -j- ~ a.- . - . -. x - avw . -?, 
dx xxx x1 
duavw dx 
u b ' x2 ' 
The reagents are present in such large excess that v and w may 
be regarded as constants (unless the permanganate suffers from 
an unusual amount of pollution), and by integration, 
. avw i , 
log« u - constant, 
. avw i . 
loge u0 =. -j- . constant, 
, u avw x. - x 
10ge = , 
w0 b 
which may be written 
. « avw - x A x. - x 
~<o = = W Warder. 
370 
This equation represents any one reaction of the kind specified 
during the concentration required in an ordinary estimation of 
albuminoid ammonia, but the mathematical expression must not 
be pressed too far. Thus, if we make x - o, this not only implies 
that the contents of the retort are evaporated to dryness, but also 
that the reagents have gained infinite concentration, and therefore 
act with resistless force, effecting complete alteration of any 
remaining traces of the nitrogenous body. The form of the curve 
varies greatly with the value of A. Two examples are given in 
the figure. Curve B represents this equation when A-i; curve 
C when A - Since the quantity of ammonia present at any 
moment (both in retort and distillate) is proportional to the weight 
of the nitrogenous body transformed, we may write 
-■ w : w0: : O + 2) : (jy + z)m, 
where subscript oo is used to denote the ultimate value theoreti- 
cally possible; or, in another form (since o), 
* - - O' + *) ( A 
' 
The curves B and C are so drawn that ordinates represent (_y -f- O 
as a function of x. 
It will be seen at once that when A - i the quantity of ammonia 
increases very rapidly during the first part of the distillation, but 
when A - o.i, (jF 4~ z) increases but slowly until the fluid is con- 
siderably concentrated. Equation (4) suggests various modifica- 
tions by which A may be altered at pleasure; for example, by 
increasing the quantity of permanganate used (represented by w), 
or by distilling a less quantity (0 each minute. Thus Smart 
noted several instances of increased action by slower boiling. 
Curves of similar properties express the progress of other 
reactions which take place during concentration in large excess of 
reagents. Equations of the following forms are obtained by vary- 
ing the exponent of x in equation (3) and using the appropriate 
constants. For reactions involving the concurrence of two mole- 
cules only, 
-loga + z) (6) 
-*0 
For three molecules, 
-log^~(j, + a') (7) 
x z 
A. Curve of distillation of free ammonia, a, a, a. Fractions, each the original volume. B. Formation of albuminoid 
ammonia, eq. (7); -4 = 1. G. Formation of albuminoid ammonia, eq. (7); D. Ammonia in distillate, from curve B. 
d, d, d. Fractions, each | the original volume. E. Ammonia in retort, from curve B. Values of x, although positive, are measured 
towards the left, in order that time may be counted towards the right. 
cz 
A 
Explanation op the Figube. 
k 
M 
- 
0 
-4 
L/ 
r 
0 Dynamical Theory of Albuminoid Ammonia. 
37i 
For four molecules, 
- log + = A *=£-. (8) 
2<o x 
In all such cases the formation of ammonia may take place 
most rapidly in the early, middle or latter part of the distillation, 
according to the value of A. 
If the ammonia, instead of appearing directly, is the final pro- 
duct of a succession of different reactions, a complete theory must 
take account of the gradual formation and disappearance of the 
intermediate products. Some evidence on this point will appear 
in section VI. 
IV. Distillation of Albuminoid Ammonia. 
Equation (i) applies to the distillation of ammonia only when 
the total quantity (y + z) remains constant; otherwise we must 
write - dz in place of dy. Thus, 
(9) 
dx x 
x dz , x 
or = -T'-2F- <Io) 
Substituting in equation (7), 
. / . xdz z\ . Xq - x . 
This equation (as soon as a fixed value is given to A) expresses 
a perfectly definite relation between the ammonia distilled and the 
volume remaining in the retort. It is not easy, however, to obtain 
an integral in form suitable for computation, and an indirect method 
was used. The ordinates of curve B represent the sum of y 
and z when A - i; these ordinates were divided into two parts, 
in accordance with the condition implied in equation (9). Curve 
D thus represents the gradual increase of z, and curve E shows 
the increase and subsequent decrease of y. A few corresponding 
points of these three curves are also indicated in the following 
table, increments of z being placed under the symbol dz. These 
increments are shown on the figure as d, d, d. For convenience, 
zm is taken as unit of weight for ammonia, and xa is made equal 
to 8; as when a volume of 400 cc. is distilled in fractions of 50 cc. 
each. The ratio of each increment to the next is stated under r. 372 
Warder. 
> + «• 
y- 
z. 
As. 
r. 
8 
.ooo 
.000 
.000 
... 
7 
.280 
.129 
•151 
•151 
•58 
6 
•535 
.122 
•413 
.262 
1.05 
5 • 
•749 
.087 
.662 
•249 
I-3I 
4 
.900 
•047 
.853 
.191 
i-74 
3 
.978 
.015 
•963 
.110 
3-23 
2 
•999 
.002 
•997 
•034 
... 
O 
1.000 
.000 
1.000 
... 
When A = i (as in this table), y reaches a maximum when 
x-£>y, during the collection of the second fraction of distillate; 
and this fraction contains the largest quantity of ammonia. The 
third fraction contains nearly the same amount, but later incre- 
ments show a rapid decrease. The large increase of ammonia 
indicated for the second measure is due in very small part to the 
increased concentration of the permanganate, but chiefly to the fact 
that sufficient time has elapsed for the accumulation of ammonia in 
the retort. An increase in the second measure has been rarely 
observed in practice, except in a few experiments on pure chemicals, 
such as urea, theine, glychocholic acid, and notably in cyanuric 
acid. In the 59 samples of natural waters already referred to, 
the first measure always contained as much as ii times that found 
in the second; in 11 instances the ratio is 2, and in 33 instances it 
is more than 2. On the other hand, instead of an increase in 
this ratio for succeeding portions (as indicated by theory for a 
single reaction), it usually decreases, the ratio of the last two deter- 
minations being from 1 to 2. This want of agreement between the 
theory for a single reaction and observations on a natural water are 
exactly what we should expect; for the water may contain a great 
number of different substances, each undergoing change accord- 
ing to equation (6), (7) or (8), so that the actual rate of distilla- 
tion depends upon the sum of the ammonia derived from all these 
sources. As curve E shows a rapid rise from zero to a maximum, 
followed by a gradual decrease to zero, so we may conceive a 
multitude of similar curves, with their several maxima at various 
points, but with a large share of them near the origin; implying 
z4)> 1, for the most part. The resultant of all such curves might 
have a single maximum, like the type given, or it might have a 
succession of maxima and minima. On the theory of averages, Dynamical Theory of Albuminoid Ammonia. 
373 
the former seems more probable. This resultant curve might have 
a great variety of form in details, answering to the various behavior 
of different waters. 
V. Exterpolation Formula for Residue in Retort. 
Many chemists, feeling that the task is not completed by ness- 
lerising three measures, have pushed the distillation much farther. 
Mallet1 suggests a certain limit as criterion when the distillation 
has been carried far enough. If the ammonia found in the suc- 
cessive portions is reported, we can rudely plat the curves for 
ammonia distilled, ammonia in retort, or total ammonia formed. 
Each of these curves may be the resultant of many similar curves 
having different parameters. Can we find an empirical equation 
for the resultant that will conform to the analytical results, and 
thus enable us to apply a probable correction for the ammoniti 
that would be obtained if the distillation were completed ? 
We may assume a curve of the form 
y - Ax Bx1 (12) 
to indicate the varying quantity of ammonia in the retort under 
the condition that this is all distilled away when the fluid is gone. 
Substituting in equation (9), 
- dz 
-A± - kA-\-kBx. 
dx 
Integrating between the limits zY and z2, 
z2 - Zi=. kA (jTj - -|- %kB (x? - x'); 
or, for brevity, 
4xz - kAArx 4- \kBAxx2. (13) 
In like manner, for a second increment of z and for the unde- 
termined portion, 
A2z = kAdtx tx2t 
dmz = kAdxx + ikB4ax2. (14) 
Those who follow Wanklyn's directions for 500 cc. of water 
begin the distillation of albuminoid ammonia with 350 cc. of fluid, 
and distill three measures of 50 cc. each, rejecting the last 200 cc. 
Taking 50 cc. as unit of volume, xY = 6, = 5, and = 4, 
while and d2z represent the ammonia found in the second and 
third measures. Assume = rd2z, then 
1 Loc. cit. p. 208, 374 
Warder. 
r42z =z (6 - 5) + hkB (tf> - 25) = kA + 5.5/L# , 
J22 - kA (5 - 4) + hkB (25 - 16) - kA -|- 4.5^2?. 
By elimination, 
kB - (r - 1) d2z and kA = (5.5 - • 
But z - \kA 4- SkB, 
= (14 - (15) 
If r= 1, J002' = 442'; 
that is, if the second and third measures yield like quantities of 
ammonia, the best correction we can make is to assume that the 
distillation of four times 50 cc. still remaining in the retort would 
yield four times that quantity. 
value of A becomes negative, and therefore y 
becomes negative for small values of x. This involves a physical 
impossibility, but we may meet the difficulty by treating the 
equation 
y- Ax1 + Bxz (16) 
in the same manner as equation (13), whence 
. 188 - loor . z A 
4*, (17) 
a correction which again fails if •* 
We may test these formulas by means of Smart's experiments 
with pure chemicals; but, as he began the action with 400 cc. of 
fluid instead of 350, and A2z represent his third and fourth 
fractions, with the advantage of greater distance from the beginning 
of the operation, where equation (12) is wholly inapplicable. The 
available data (for values of r between 1 and 1.5) are given in the 
following table, which states under 
1 To obtain a more general formula for the correction proposed, we may assume 
y = Ax'1 4- Bx" +1, (18) 
whence, by the same treatment illustrated above, 
<A,x-|--rA„-+1)A..A,a') A.^1 ( 
where A is positive or zero when 
, A,jr" + 1 
We may either make n the smallest integer consistent with the condition (20), or we may 
determine n by making the fraction equal to r. The function A00z = </>(r) will be a smooth 
curve in either case, consistent with the theory as developed above and with the analytical 
results. Dynamical Theory of Albuminoid Ammonia. 
375 
A, per cent, ammonia found in the first four fractions; 
B, per cent, ammonia found in " exhaustive experiments"; 
C, per cent, ammonia obtainable after the first four fractions; and 
D, per cent, ammonia indicated by the exterpolation formulas. 
A. 
B. 
c. 
D. 
Silver and potassium cyanide, 
2.25 
[i7-i] 
14.85 
2.4 
Prussian blue, 
•3 
25-8 
25-5 
.2 
Urea, from urine, 
( 8.7 
43-io 
34-4 
8.8* 
1 8.65 
34-45 
6.6 
Urea, from ammonium cyanate, . 
f 7-7 
43-4 
35-7 
8.0 
I 7-5 
... 
35-9 
8.0 
Alloxan, 
f 6.55 
21.3 
14-75 
2-5 
( 6.6 
... 
14-7 
2.4 
Thiocarbanilide, 
4-i5 
11.85 
7-7 
1.6 
Thiocarbamide, 
f 15-25 
22.50 
7-25 
i-7 
1 15-0 
... 
7-5 
1.6 
Strychnine sulphate, 
5-4 
i-4 
•4 
t 3-75 
1-65 
1.0 
Codeine, 
5-o 
5-32 
•32 
•5 
A glance at the last two columns shows that the ammonia actu- 
ally obtainable after one-half the fluid had been distilled was many 
times the proposed correction. An explanation of this discrepancy 
will next be considered. 
VI. Evidence of Intermediate Compounds of Nitrogen. 
Making some allowance for the difference between equation (7) 
(which was selected for study) and equations (6) and (8), it is safe 
to conclude from the table in section IV, that if the action of alka- 
line permanganate upon a pure compound of nitrogen could be 
represented by any single chemical equation (no matter how 
complex), we should have reason to expect this change to take 
place with such speed that fully 70 per cent, of the theoretical 
yield of ammonia would be obtained by distilling one-half the 
fluid, or else with such slowness that the first measure would 
contain far less ammonia than the second. Published experiments 
with at least eight bodies1 contradict both these conditions, and 
we are forced to believe that in the earlier stages some molecules 
of the original body yield ammonia with promptness, while others 
1 These are the first six named in the last table, besides parabanic acid and theine. 376 
Warder. 
are converted into more stable forms and resist the action with 
great stubbornness. The nature of these intermediate products is 
not yet clear, except only that a (as defined in section III) is small, 
in their reaction with alkaline permanganate. Perhaps they may 
be chiefly condensation products. Alloxan, which yielded 4 per 
cent, ammonia in the first fraction (eighth) of distillate, and only 
2.57 per cent, in the next three, but 21.3 per cent, by prolonged 
action, is a marked example of such behavior; this, no doubt, 
is partly converted into parabanic acid. So, likewise, in a water 
analysis, many substances may yield a small part of their nitrogen 
as ammonia, while the greater part is converted into more stable 
forms which elude detection. It has been supposed that the 
organic matter of potable water is more like asparagine, yielding 
its nitrogen readily under the action of permanganate. Of this, 
however, I have found no proof. In about 40 per cent, of the 
natural potable waters referred to, the ratio of the third and fourth 
fractions is 1.5 or less, giving grave suspicion of incompleted 
action when the distillation was stopped. Even if the yield of 
ammonia in some fraction has fallen below the limit of delicacy 
in nesslerising, what evidence have we that prolonged treatment 
would not develop a much larger amount of ammonia than that 
already obtained ? 
VII. Applications to the Analytical Process. 
A debt of gratitude is due to Professor J. Alfred Wanklyn, as 
the author of a valuable method of analysis which has done much 
in the interest of public health. " The ammonia process," which 
originated in 1867, is still described in the 7th edition of "Water 
Analysis" (1889) essentially as in the author's first paper upon the 
subject. In a process of such delicacy, where the results may be 
influenced by a multitude of variable conditions, it is desirable that 
these should be accurately defined and conscientiously followed. 
According to the author's latest directions, 500 cc. of water are 
reduced to 300 cc. (to remove free ammonia) and 50 cc. of the 
reagent added ; three measures of distillate are then collected, when 
" the distillation may be stopped. ... It is necessary to nesslerise 
each separate 50 cc. of distillate, and to add the amounts together, 
in order to arrive at the total albuminoid ammonia." Besides this 
process, a second is described upon a scale of one-fifth, with this 
important modification, that the operation is " continued so long Dynamical Theory of Albuminoid Ammonia. 
377 
as io cc. of distillate shows a sign of ammonia"; if, then, with a 
delicate Nessler reagent, ammonia is found in the sixth measure 
of distillate, the fluid must be evaporated to dryness! 
But why should 350 cc. be distilled until only 4 have gone over, 
while 70 cc. are distilled up to a very different limit ? Probably 
because the time required to evaporate 30 cc. from the smaller 
retort proved insufficient for both the chemical action required 
and for the separation of the ammonia. Professor Wanklyn's 
method, as conducted in his own laboratory and by his own hands, 
has doubtless done valuable service, yielding good comparable 
results; but more definite conditions must be adopted if we would 
avoid discrepancies among different places and persons. Its value 
depends upon giving constant and comparable results under fixed 
conditions. For the absolute estimation of organic nitrogen, it is 
worthless. 
It will be instructive to divide the free or albuminoid ammonia 
found in each measure of distillate by that found in the next, and 
to record the ratio so found; especially if care is taken to boil at a 
constant rate. According to Smart,1 with pure animal or vegetable 
albuminoids, r=2; but if these have passed into the putrefactive 
condition, r=3; while in the case of urea (after the preliminary 
stage is passed) r~ 1, both for free and for albuminoid ammonia. 
The successful detection of recent sewage is reported, when the 
chemical evidence consisted in the persistent evolution of free 
ammonia, followed by persistent evolution of twice as much albu- 
minoid ammonia. 
The term "albuminoid ammonia," like "reverted phosphoric 
acid," has a technical significance determined by the method 
adopted in the analysis. The actual market value of reverted 
phosphoric acid cannot be so high under the rule for extraction 
at 65° as under the rule for extraction at 40°; so also 0.1 mgm. 
of ammonia obtained by prolonged and intense action is of less 
real importance than the same quantity obtained by less vigorous 
measures. Many modifications of the original process have been 
proposed; among these Breneman's2 are worthy of special con- 
sideration. If the fluid is brought nearly to dryness, as he suggests, 
it may be possible to apply an exterpolation formula, such as 
equation (19), to the measured residue. The detailed record of 
1 Public Health (Minn.), Vol. JI, No. 3, May, 1886. 
2 Jour. Amer. Chem. Soc. 8, No. 9. 378 
Warder. 
ammonia found in seven successive fractions (with greatly increased 
temperature and concentration at the last) may help to distinguish 
between different kinds of nitrogenous matter, yielding ammonia 
more readily and less readily; but care should be used in regard 
to the volume of the residue; if this is 4 cc. in one case and 7 cc. 
in another, there may be considerable difference in the oxidation 
of that comparatively inert matter which we have been led to 
suspect. 
If the ammonia process should be applied to investigations 
of the relative stability of organic compounds,1 the formulas 
given above may be adapted to numerous possible modifications. 
The use of return-flow condenser (to prolong the chemical action 
apart from distillation), and variations in the proportions of potash 
and permanganate, do not seem to have yet been tried. It is 
difficult to see why the alkali should always be used in so much 
greater excess than the oxidiser (in spite of inconvenient bumping), 
but it is especially strange that the distinguished author, ignoring 
Mallet's report of 1882, and republishing his directions without 
any revision, should leave the important question of time entirely 
out of account. " Suffering persecution for the sake of the truth " 
should make the scientist keenly alive to any truthful criticism, 
and ready to study the contributions of his colleagues. The 
"ammonia process" has now passed from one laboratory into a 
hundred. If "discovered about a generation before chemists 
were prepared for it," the time is now fully ripe to collect the 
testimony of those who have used it; to agree upon the general 
conditions of quantity, rate of boiling, and limit of distillation, and 
even upon such details as the matter of paper or rubber connec- 
tions. When a satisfactory compromise is effected we may hope 
to have a uniform method of procedure and a real definition of 
albuminoid ammonia. 
Washington, D. C., August, 1889. 
1 It may be of interest to compare such results with those of Dreyfus, who studied the rate 
of oxidation with acid solution of permanganate. Compt. rend. 105, 523.