AmiiS'  1913™' }    Standardization  of  Digitalis  Bodies.  103 
last  animal  being  abnormal  in  reaction.  It  is,  therefore,  obvious 
that  one  must  not  trust  the  reaction  of  a  single  animal,  or  take  the 
average  of  the  results  of  tests  on  two  animals  which  show  con- 
siderable variation.  In  view  of  this  fact  one  should  discard  those 
results  which  were  obtained  upon  animals  showing  obvious  abnor- 
mality of  response.  Such  a  practice  is  common  to  all  of  the  methods 
of  assay  here  considered  and  is  an  entirely  justifiable  procedure. 
Applying  this  means  of  rendering  the  use  of  animals  more  exact, 
from  a  quantitative  point  of  view,  to  the  series  under  discussion, 
one  would  have  to  discard  the  three  starred  experiments  before  mak- 
ing his  calculations.  The  average  dose  calculated  from  the  remain- 
ing nine  tests  is  0.100  mg.  per  kg.  of  cat,  which  is  precisely  the 
average  previously  determined  for  this  specimen  of  ouabain.  With 
this  correct  average  and  the  three  very  abnormal  results  discarded 
there  is  still  a  maximum  variation  from  the  average  of  16  per  cent, 
in  the  case  of  one  experiment.  This  may  be  explained  on  the  ground 
of  season,  a  matter  to  be  discussed  subsequently. 
When  the  same  arguments  and  methods  of  correction  are  ap- 
plied to  some  of  the  other  series  we  see  how  much  greater  is  the 
accuracy  of  the  cat  method  than  was  at  first  thought.  Thus,  in 
series  B  of  ouabain  experiment  2  was  obviously  made  on  an  abnor- 
mal cat  and  when  discarded  from  the  calculation  the  variation 
between  the  three  experiments  remaining  falls  to  3.8  per  cent,  of 
the  average.  Again,  in  the  series  with  helleborein,  the  figure  2.47 
is  obviously  far  too  high  and  the  animal  from  which  it  was  obtained 
was  certainly  tolerant.  With  it  eliminated  the  maximum  variation 
from  the  corrected  average  is  only  1.7  per  cent,  of  the  average. 
I  might  here  state  that  in  a  series  of  standardizations  of  24 
different  specimens.,  including  many  different  types  of  digitalis 
bodies  and  extending  over  nearly  two  years,  variations  were  found 
ranging  from  less  than  one  per  cent,  up  to  17.4  per  cent,  of  the 
average  taken  in  each  series.  The  average  variation  was  6.1  per 
cent,  for  the  entire  24  series  of  tests. 
In  carrying  out  the  cat  method  of  standardization  in  actual 
practice  at  least  three  animals  are  used  and  if  all  thjjee  give  closely 
similar  results  the  average  is  taken  as  being  correct.  If  two  of  the 
results  are  quite  close  but  the  third  is  considerably  at  variance  a 
fourth,  or  even  a  fifth,  test  is  made  before  an  average  is  struck. 
An  exception  to  the  use  of  three  tests  is  sometimes  made  when 
the  first  two  are  found  to  give  very  closely  similar  results,  as  for 
