428  Stability  of  Digitalis  Leaf  Extracts.  {Am-^\^m- 
water  and  when  this  extract  has  been  exhausted  with  ether  and 
chloroform,  a  very  considerable  part  of  the  glycosides  remain  solu- 
ble in  water  and  not  in  chloroform  or  ether.  Here  is  a  part  of  the 
subject  even  less  understood. 
Keller10  attempted  to  duplicate  Kiliani's  work  on  a  small  scale 
(20  Gm.)  and  obtain  a  quantitative  estimation  of  "  digitoxin  "  and 
thus  arrive  at  a  conclusion  concerning  the  activity  of  the  sample. 
The  leaves  were  extracted  with  70  per  cent,  alcohol,  percolate  con- 
centrated, precipitated  with  lead  acetate,  filtered,  and  excess'  of  lead 
removed  with  sodium  sulphate.  The  clear  fluid  is  rendered  ammo- 
niacal  and  extracted  with  chloroform.  These  extracts  are  concen- 
trated, redissolved  in  3  Cc.  chloroform  and  precipitated  with  an 
ether  -f  ligroin  mixture.  This  precipitate  is  weighed  and  calcu- 
lated as  digitoxin.    Yield  0.26-0.62  per  cent. 
To  illustrate  the  futile  nature  of  Keller's  assay  we  give  an  ex- 
periment wherein  the  activity  of  the  "  crude  digitoxin  "  amounted 
to  only  12  per  cent,  of  that  of  the  drug. 
A  tincture  of  digitalis  U.S. P.  (D-1625)  was  assayed  by  the  one-hour 
frog  method.    Found  M.S.D.  =  0.0052  Cc. 
Assayed  by  Keller's  process  the  total  crude  chloroform  extract  (12  Gm. 
drug)  weighed  73,1  Mg.  =  0.6  per  cent.  Redissolved  and  precipitated  with 
ether  -f-  ligroin  the  "crude  digitoxin''  weighed  48.4  Mg.  (0.4  per  cent.). 
This  was.  dissolved  in  200  Cc,  70  per  cent,  alcohol.  Forty  Cc.  (2.4  Gm. 
drug),  was  concentrated  on  the  steam  bath  to  remove  alcohol,  and  then  made 
up  to  20  Cc.  with  physiological  salt  solution.  (Solution  =  %  X  tincture 
strength.)  M.S.D.  =  0.0366  Cc.11  The  proportionate  toxicity  of  Keller's 
"  crude  digitoxin  "  is  therefore  about  12  per  cent,  of  the  total  of  this  drug. 
We  did  not  carry  this  work  any  further,  as  it  has  already  been 
shown  by  Barger  and  Shaw12  that  the  digitoxin  isolated  in  Keller's 
process  represents  from  29  per.  cent,  to  14.5  per  cent,  of  the  activity 
of  the  drug.  This  has  also  been  established  by  Focke  and  others, 
and  the  proof  is  established  that  the  chemical  process  above  de- 
scribed is  entirely  useless. 
The  interest  in  Keller's  work,  however,  accentuates  the  value  of 
pharmacological  work  to  chemical  investigation.  It  is  evident  that 
a  large  part  of  the  active  substance  or  substances  are  either  (a)  lost 
in  the  resinous  precipitate  and  lead  acetate  precipitate  or  (b)  con-, 
10  B.  d.  d.  Pharm.,  G.  7,  125  (1897)- 
11  Equivalent  to  0.018  Mg.  p.  Gram  frog  Keller's  "  digitoxin "  i.e.,  0.5 
Mg.  for  30  G.  frog. 
12  Pharm.  J.,  73,  259  (1904)- 
