AMMONIA AND NITRIC ACID IN RAIN-WATER. 
167 
•bout 35i ounces) of water, and collecting and testing by the alkalimetrical 
method, two or more successive fifths or tenths of the product. or 11 s 
■'Jrpose M. Boussingault uses a test acid little more than £th the strengtli ot 
wracid at 10°, and only JUth the strength of the test acid ot M. 1 eltgot. 
Nb test alkaline solution, again, is not quite ^rd the strength, volume tor 
-ilume, or his test acid. In practice he finds that the limit ot error in tie 
'-*nf this dilute alkaline solution is about 0*2 cub. cent, measures of it, equa 
to about 0*083 milligramme, equal to 0*0005 grain of ammonia. This, equal to 
1 '"“tjitayth of a grain only, is certainly a surprisingly small actual amount 
'ammonia to determine by analysis; hut when it is considered, that in the 
entity of water operated upon by M. Boussingault, there is seldom 1 mil- 
'Sfarnme= 0 ‘ 0 i 54 grain of ammonia, it is obvious that an error ot on y 
^ milligramme, or 0-005 grain, however small in actual amount, is still 
n«iilerable in relation to the whole to be estimated, amounting in the case 
Opposed to about »UJi or aird per cent. But when we further consider, that 
t4,ta l amount of ammonia in a litre of water is frequently considerably less 
in » milligramme, and also that this total amount, whatever it may be, is 
T- fur testing into two or more separate portions of the distillate, it is 
. ! ou * tllat the minimum range of error, especially in the testing of the 
'*“ !r Portions, must bo very considerable indeed in proportion to ita whole 
uu t. We have however 1 had the privilege of witnessing the conduct of 
)j | > . rocess in the hands both of M. Boussingault and of his able assistant 
I(iT U | Z , ea »'. atlt l cer tainly with surprising uniformity of result. 
Nil able IL ar e given the actual results of experiments in the Rothainsted 
•nbrnfi I?‘ whicb 60 ounces, instead of only 1 litre, were generally 
^ distillation ; ami although the greatest care was taken, it must 
•■flminn pi these first results of uii practised hands, neither was the 
»( l “ e ammonia in the successive portions of the distillate so uniform 
Limoni! If „ experimen * 3 M- Boussingault, nor were the total amounts ot 
lislip,! i 0Un(1 >n duplicate specimens of water so coincident as could be 
point I,’.* . t , lt mu st be clearly understood, that attention is called to this 
M.p, ,‘V n he leii «t with the view of depreciating the admirable labours of 
wb ‘ cb wc arc satisfied have been conducted with the ntmos 
knr, e l. whicI > tire process is capable, but it seems desirable to point out 
tar e f ul bc ^ proportional error in less constantly practised or less 
Pic.-ptini!''!'* J(? mont, oned, that in our conduct of this Method 2, ill all cases 
*- lhc month oF April, a test acid of 1° strength only was em- 
‘"■ireo «L la V s t0 sa y. ^tli the strength of that used previously, and it was 
•llmli, on y about fills the strength of Boussingault’s acid. I ho test 
* 1-1 aftBMuaJ other hand, was, for the rain of the first four months, at 1 , 
U,fc *arn c at °«ly 1° strength, which latter is very nearly identically 
!‘ r a semen,? ? ed b > Boussingault. We sought to measure too to a *th 
bf |ior, w i.j i° ”*e stronger, and J a septem of the weaker alkaline test- 
^ anin l0 i,h * . ros P Pct ' v ‘ily represented 0034 milligramme, or 0’00054 grain 
• B«u * • wt MehtleWly the same degree of accuracy as that attained 
^tnino ot . SSin S ft «lt in measuring to 0*2 cub. cent, of his alkali =0‘0iS milli- 
i*^ s , narkpi| aim ° ,1 ' a * The fractional distillates were collected in small 
til| e( j acc °rdmg to the quantity to be collected, each of which, 
^ Un ' nivasur • Corkc<l . U I' unfit fire series was ready for testing. An 
with th„ j. P.i 1Pe distilled water was then put into a te6t-glass side by 
V? acid as » '^bHates, and to it was added the same measure oflitmus 
^'valent of ti * 16 ra * n products. This being neutralized with its exact 
re test-alkali, furnished a guide as to the tint to be aimed 
