1014 
with  the  addition  that  the  impregnation  in  a thirty  per  cent.  AgNO, 
solution  was  made  specially  long  — over  ten  days  and  nights.  The 
impregnated  pieees  were  imbedded  in  paraftin  and  were  cut  into 
conveniently  Ihick  sections,  5 — 15  p.  A nnmber  of  spinal  ganglia 
were  cut  in  unbroken  series  of  sections,  10 — 15  in  thickness. 
Cross  sections  wei'e  placed  on  the  sub-dnral  parts  of  the  spinal 
nerves,  partij  close  to  (centrally  of)  (he  spinal  ganglia  and  partly 
close  to  the  spinal  cord.  In  the  [)reparation,  in  which  the  spinal  cord 
and  the  spinal  ganglia  had  been  fixed  in  sitii  in  the  canalis  nied. 
spinalis,  cross-sections  were  ent  right  from  the  caudal  end  as  far 
in  the  direction  of  the  cranium  as  the  sub-dural  part  of  the  nerve 
roots  had  a caudal  course  — this  was,  as  a rule,  up  to  the  posterior 
third  and  the  posterior  half  of  the  thoi-acal  veitebral  column.  The 
rest  was  cut  into  sagittal  sections,  during  which  the  microscope  was 
used  to  verify  that  cotneniently  sitiiated  parts  of  the  segmental 
nerves  were  present  in  the  sections.  These  sections  were  made 
5 — 10  p thick.  In  determining  the  mimber  of  nerve  übres  in  the 
cross  sections  1 used  a Leitz  microscope  (tripod  Gr.H.)  with  a 
cross-table,  an  oil-immersion  7n  ocular  IV  (Leitz) 
with  the  enclosed  sqiiared  glass  plates.  It  appeared  to  be  necessary 
to  work  with  such  a great  magnifying  power  in  determining  the 
number  of  nei’ve  fibres  in  order  to  be  able  to  disin tegrate  those 
parts  of  the  preparation  in  which  the  nerve  fibres  were  most 
close,  especially  in  the  yonng  animals.  Before  beginning  to  count, 
the  square-ocular  and  the  cross-table  were  adjusted  so  as  to  prevent 
as  far  as  possible  unexiiected  displacements  and  miscalculations 
arising  from  these.  Repeated  calculations  with  the  same  preparations 
have  also  shown  that  the  errors  in  calculation  that  we  are  concerned 
with  are  small  — always  less  than  ten  per  cent,  as  a rnlenotraore 
than  üve  per  cent.  At  first  I attempted  an  approximate  method  in 
deciding  the  niimber  of  nerve  fibres  in  the  cross-sectioned  nerve 
roots.  1 counted  each  nerve  fibre  in  a few  hundred  squares  and 
found  the  average  number.  I then  counted  the  number  of  squares 
in  a cross-section  and  multiplied  this  by  the  average  number.  This 
method  appeared,  however,  to  give  values  that  were  too  uncertain, 
because  nerve  fibres  of  different  thicknesses  were  very  unevenly 
distributed.  In  order,  therefore,  to  obtain  siifficiently  exact  values, 
I was  thus  compelled  to  count  every  nerve  fibre  in  the  whole  cross- 
section  — a method  that  was  certainly  trou blesome,  but  necessary 
in  this  case,  especially  with  young  animals.  In  connting  1 always 
began  at  the  top  and  at  the  left,  both  in  the  preparation  and  in 
the  field  of  the  squares,  taking  care  that  all  the  nerve  fibres  which 
