170 
REPORT—1854. 
A glance at the columns of detail of this Table clearly illustrates the fact 
pointed out by M. Boussingault, upon which this method of determining the 
ammonia in rain and other waters is based, namely, that in the case of very 
dilute solutions of the volatile alkali, nearly the whole of it passps over in the 
first portions of the distillate. Wo observe, however, that even where the 
“ total amount ol ammonia obtained from duplicate specimens of water is 
the same, or very nearly so, still the quantities determined in the correspond¬ 
ing fractional parts do not always agree so well. This would seem to indi¬ 
cate that the discrepancy is due to irregulurity in the distillation, rather than 
to error of observation in the testing. It ia true, that this method of suc¬ 
cessive distillations is much more tedious and troublesome than the method 
ol small single distillations preferred by M. Boussingault; ami even with it, 
it would appear, as in one or two cases in Table III., that the difference 
between the amounts of ammonia determined in duplicate specimens of rain- 
water, may in practice be nearly as great in proportion to the whole, as that 
which is assumed to he a not necessarily exceeded range of orror in tin; 
method of small single distillations. Upon the whole, however, we consider 
the modification involved in this Methods, to be practically very important, 
and that this form of the process of estimation by distillation is much more 
applicable to this delicate subject of inquiry than either of those which we 
have previously adopted. 
Before leaving the question of method, it may lie well to enumerate the 
calculated minimum proportions of error by the different methods, supposing 
r ”}‘ 1,1 * et “°ds2and3 equally with Method 1, this error would necessarily 
y occl ‘r ® n . oe 1,1 a f otal estimation. This however is not the case; for it is 
r r:rV n ?- raCt T, U n ,' i « ht tend ^ the same direction in the estin.a- 
. h \\ Ik-, !i ^ distillate. It is on the other hand of course pos- 
fractiong ^ ,?rror ’ 1 ,lcx ' 8t ®d> might be counterbalanced among the several 
nrod^r«f^inA and supposing one-tenth of the concentrated acid 
nf fl,.. , 00 ll , rCS 0 water to be finally tested, this would, on the average 
0 -] r.iq o r U ° f f< .' UU ! * u ra ' u a t Kothamsted, contain 10 milligr-— 
tier of,?, .!! ?[ an , lllln nia. Assuming also, as was found in the prac- 
the arid i l |t 0 r’ thal t ‘ ,e degree of accuracy easily attainable in the use of 
of ainn.,1, * ? nt kSt ,ir l Uord at indicated 0*2 niilligr.=U*00Sl g™" 
ccnt of Ul i ra ' ,KC 0f error woald obviously amount to tfk, or 2 per 
be borne in lo a "* niom a contained in the product tested : and it should 
difficult v w,“r (K ‘f far as the Minffw as concerned, there was no 
together • ,l.. .Vi" " lj,uiriui 8 duplicate estimations which agreed absolutely 
practice than ihnt lcanioaut error in a scries was certainly nol more in 
as before , The difficulty with this method however was, 
vessels. the mana g 0fIj ent of such large distillations in gla- 4 ® 
posed to be di. t,lat j f Vr ’ I,0,,R »>nfiault, 0*033 milligr. of ammonia are sup- 
Tlie uverie . *. nd tll<! arru,, ‘nt of water operated upon to be 1 h' rr * 
HotliamaiefH °L atnn,or, ia in I litre of the rains collected at 
error w ould dT • U V V" *'gr., or 0*0154- grain ; so that the minimum limit o 
The aveSi 0 ^' 0 ^ V 3 P" <*"*• of the whole amount in such a case, 
the rain ^the^n^*^ ° f amrnonia I»?r litre found by M. Boussingault in 
mhiimmu enlT T**? ' ,OCa Dot exc ^ J milligr.; in which case the 
rather more »i,„n a * ' lluatin 8 to 0*033 milligr. would amount to 0 
amount of aniuioni* P ft*S2’ ° f t,1U who,e * 1,1 the water of rivers, the average 
per litre • umm u l • \ ® 0r dmg to M. Boussingault, was not quite 0*2 nll ^'8 * 
P hire, upon which, determining only to the same amount as before, the 
