104  Standardisation  of  Digitalis  Bodies.    { AmM^ch'  2™" 
example,  in  the  cases  of  apocynum  and  the  tincture  of  strophanthus 
cited  in  the  table.  Even  in  such  circumstances,  if  the  test  is  one 
of  a  preparation  for  use  in  man,  or  of  one  which  is  wholly  unknown 
as  to  approximate  cat  unit,  a  number  of  tests  is  always  made  for  the 
sake  of  greater  accuracy  and  certainty. 
It  may  be  contended  that  two  of  the  three  or  four  cats  which  are 
employed  in  a  routine  assay  might  readily  be  abnormal  in  reaction, 
and  in  the  same  direction.  This  would,  to  a  certain  extent,  be  an 
invalidating  feature  of  the  method  if  such  an  occurrence  were  likely, 
but  it  is,  fortunately,  an  almost  impossible  accident,  for  it  has  not 
been  possible  to  find  an  abnormal  reaction  of  thirty  per  cent,  or 
over  in  more  than  eight  animals  in  300;  that  is,  only  2.7  per  cent, 
of  all  cats  are  found  to  be  very  abnormal  in  reaction.  This  group 
of  300  experiments  is  made  up  by  beginning  at  the  latest  series  and 
counting  back  for  nearly  two  years,  including  every  experiment 
in  each  series  of  tests.  The  figure  is  therefore  a  fairly  trustworthy 
index  of  the  frequency  of  occurrence  of  decidedly  abnormal  animals. 
No  evidence  is  to  be  found  in  this  series  to  show  that  either  tolerance 
qr  susceptibility  is  the  predominant  abnormality,  hence  the  chance  of 
two  animals  falling  in  a  single  series  and  having  the  same  type  of 
abnormality  of  reaction  is  very  slight  indeed.  Just  such  a  rare 
coincidence  was  observed,  and  is  shown  in  series  A  of  the  ouabain 
tests  given  in  the  table,  experiments  5  and  9.  This  was  a  long 
series  and  the  three  abnormal  experiments  present  are  seen  to  have 
been  scattered  throughout  the  group,  so  that  had  any  set  of  four  or 
five  tests  been  made  these  abnormal  reactions  would  have  been 
recognized  as  such  and  due  allowance  made  for  them. 
Comparing  the  variations  encountered  in  the  cat  method  with 
some  of  the  reported  results  of  the  frog  methods  of  testing,  or  with 
the  guineapig  method,  we  find  that  6  out  of  60  frogs  showed  an 
error  of  from  9  to  10.6  per  cent.5  Hale  asserts  6  that  the  frog 
method  gives,  "an  estimate  of  the  amount  of  glucosides  in  a  solution 
to  within  a  few  (1-10)  per  cent,  of  absolute  accuracy."  But 
Vanderkleed  7  has  shown  that  with  uniform  conditions  the  lethal 
dose  of  ouabain  for  female  frogs  is  about  19  per  cent,  greater  than 
for  males. 
How  these  variations  compare  with  the  results  obtained  by  the 
5  a.  j.  pharm.  ;  191 1 ;  83;  p.  97. 
6Hygien,  Lab.  Bull.;  No.  74;  191 1. 
7  a.  j.  pharm.;  1912;  84;  p.  14. 
