^  May^^glS:}       When  Is  Poison  Not  a  Poison?  323 
cases  where  we  can  take  the  valence  of  nitrogen  to  be  3,  oxygen  and 
sulphur  2,  chlorine  and  hydrogen  i.  In  other  cases,  such  as 
those  where  quinquivalent  nitrogen  has  been  assumed,  the  new 
theory  gives  results  different  from  the  old,  but  in  each  case  in  better 
agreement  with  the  facts. 
The  theory  indicates  a  series  of  new  relationships  between  cer- 
tain types  of  substances  which  I  have  termed  isosteric  substances. 
For  example,  it  indicates  that  the  molecules  of  carbon  dioxide  and 
nitrous  oxide  should  have  nearly  identical  structures  and  this  is 
borne  out  by  the  extraordinary  similarity  in  the  physical  proper- 
ties of  these  gases.  Nitrogen  and  carbon  monoxide  constitute  an- 
other pair  of  gases  which  are  similarly  related.  The  same  theory 
also  points  out  a  number  of  previously  unsuspected  cases  of  simi- 
larity of  crystalline  form  (isomorphism). 
It  is  clear  that  in  the  past  the  term  valence  has  been  used  to 
cover  what  we  may  now  recognize  as  three  different  types  of  valence, 
as  follows: 
1.  Positive  valence:  the  number  of  electrons  an  atom  can  give  up. 
2.  Negative  valence :  the  number  of  electrons  an  atom  can  take  up. 
3.  Covalence:  the  number  of  pairs  of  electrons  which  an  atom 
can  share  with  its  neighbors. 
It  is  recommended  that  only  for  valences  of  the  covalence  type 
should  definite  bonds  be  indicated  in  chemical  formulas.  One  of 
the  particular  advantages  of  the  present  theory  is  that  it  becomes 
easy  to  distinguish  between  covalence  and  the  other  types  and  thus 
to  predict  with  certainty  in  what  way  electrolytic  dissociation  will 
occur,  if  at  all. 
References. 
W.  Kossel,  Ann.  Physik,  49:  229,  1916. 
G.  N.  Lewis,  /.  Am.  Chem.  Soc,  38:  762,  1916. 
Langmuir,  /.  Frank.  Inst.,  187:  359,  1919;  /.  Am.  Chem.  Soc,  41:  868, 
i543»  1919;  42:  274,  1920;  Proc.  Nat.  Acad.  Sci.,  5:  252,  1919. 
WHEN  IS  POISON  NOT  A  POISON?*,^ 
By  John  Uri  LivOyd,  Pharm.  M., 
cincinnati,  ohio. 
A  reply  to  the  above  question  might  be:  When,  for  any  reason, 
a  substance  is  harmless  in  action,  immediate  or  remote,  be  it  applied 
