318  New  Determinations  of  Ammonia,  Chlorine,  and 
in  the  colder  months  (October  to  January)  of  the  period,  we 
find  that  the  deposits  below  0'04  inch  show  on  an  average  a 
considerable  preponderance  of  ammonia  in  the  warmer  months, 
while  several  of  the  small  groups  of  heavier  rainfalls  show  a 
preponderance  of  ammonia  in  the  colder  months.  That  the 
atmosphere  is  far  richer  in  ammonia  in  summer  than  in  winter 
is  undoubted : but  it  by  no  means  follows  that  a summer 
rainfall  should  always  contain  more  ammonia  than  a winter 
rainfall  of  similar  amount,  both  occurring  under  similar  condi- 
tions as  to  previous  washing  of  the  air.  If  the  atmosphere  was 
at  the  same  temperature  at  both  seasons  this  would  indeed 
necessarily  be  the  case,  but  with  a fall  of  temperature  the  rain 
becomes  capable  of  taking  up  a larger  proportion  of  the  atmo- 
spheric ammonia.  The  proportion  of  ammonia  in  rain  should 
theoretically  be  highest  when,  by  a sudden  change  of  tempera- 
ture, a cold  aqueous  deposit  is  formed  in  a previously  warm 
atmosphere ; such  conditions  occur  with  changes  of  wind  from 
south  to  north,  or  when  dew  is  deposited  on  a clear  night 
following  a warm  day. 
The  influence  which  the  direction  of  the  wind  has  on  the 
proportion  of  ammonia  present  in  rain  can  only  be  very  par- 
tially studied  from  the  data  before  us.  The  difference  in  the  in- 
fluence of  warm  and  cold  winds  is  apparently  similar  to  that 
already  noticed  as  characterising  the  seasons  of  summer  and 
winter.  The  smaller  aqueous  deposits  are,  as  a rule,  dis- 
tinctly richer  in  ammonia  with  a south  wind,  but  the  same 
effect  is  not  perceptible  in  the  case  of  the  heavier  rainfalls. 
Dr.  Frankland  had  called  attention  to  the  possible  contamina- 
tion of  the  rain-water  from  impurities  on  the  surface  of  the  rain- 
gauge.  During  the  present  series  of  determinations  all  bird- 
marks  were  wiped  off  as  soon  as  noticed,  their  occurrence  being 
at  the  same  time  recorded.  It  is  difficult  to  perceive  any  con- 
nection between  these  marks  and  the  proportion  of  ammonia  in 
the  rain.  Without  going  into  detail,  it  may  be  mentioned  that 
these  marks  were  most  abundant  from  August  1 to  8,  and  that 
during  the  whole  of  September  and  December  not  one  was 
noticed. 
In  Table  II.  will  be  found  a summary  of  the  results  obtained 
from  the  analyses  of  the  daily  samples  of  rain  ; the  rainfalls  are 
grouped  according  to  quantity.  The  average  composition  of 
each  group  of  rainfall,  and  the  general  average  at  the  foot  of  the 
Table,  are  calculated  so  ajs  to  represent  the  composition  of 
the  waters  if  mixed  together.  Thus,  if  the  whole  rainfall  of  the 
period  had  been  united  in  one  volume,  it  would  have  contained 
0 248  of  nitrogen  as  ammonia  per  million  of  water.  The  other 
average  compositions  given  in  this  Report  will  be  calculated  in 
