322  Mr.  S.  A.  Shorter  on  the 
glass,  and  had   a  diameter  of  15*50  cms.     The  coefficient  of 
surface-rigidity  was  calculated  from  the  formula 
n  = 
The  omission  of  the  term  involving  the  logarithmic  decrement 
is  admissible  even  when  the  damping  is  excessive. 
The  vessel  was  placed  on  a  graduated  turntable,  and  the 
disk  was  set  into  motion  by  sharply  turning  the  table  through 
about  half  a  degree.  The  oscillations  were  timed  by  observing 
the  end  points  since,  of  course,  the  centre  of  oscillation  was 
indeterminate.     A  stop-watch  reading  to  ^  sec.  was  used. 
Method  of  Experiment. — The  disk  was  cleaned  by  polishing 
in  the  lathe  with  fine  emery  and  afterwards  wiping  with  a 
clean  cloth.  The  glass  vessel  was  cleaned  by  treatment  with 
hot  chromic  acid,  followed  by  repeated  washings  with  distilled 
wrater.  The  disk  was  suspended  and  allowed  to  attain  a  state 
of  steady  oscillation.  The  vessel  was  placed  centrally  on  the 
turntable  and  adjusted  concentric  with  the  disk.  The  solu- 
tion was  then  run  in  from  a  large  separating  funnel — a  little 
having  previously  been  run  to  waste  in  order  to  ensure  a 
fresh  surface — care  being  taken  to  avoid  the  formation  of 
bubbles  on  the  surface.  During  this  operation  the  motion 
of  the  disk  was  observed  through  the  telescope. 
Six  solutions  were  used,  containing  one  part  of  saponine 
in  102,  103,  10\  105,  106,  107  parts  of  water  respectively. 
In  the  case  of  the  first  three  solutions,  the  slow  oscillatory 
motion  was  stopped  immediately  the  solution  came  into  con- 
tact with  the  disk.  Before  coming  to  rest  the  disk  generally 
executed  a  few  rapid  oscillations  of  small  amplitude.  These 
oscillations  could  be  set  up  by  rotating  the  turntable.  In  the 
case  of  fresh  surfaces  it  was  never  possible  to  observe  more 
than  four  oscillations,  so  that  the  time  of  oscillation  could 
only  be  measured  very  roughly.  Different  experiments 
carried  out  with  these  three  solutions  gave  for  the  initial 
period  values  ranging  from  2"2  to  2  5  sec,  corresponding  to 
absolute  values  of  the  surface-rigidity  varying  from  83  to 
65  dynes  per  cm.  This  variability  is  not  surprising  when 
we  consider,  first,  that  it  is  difficult  to  measure  the  time  of 
oscillation  accurately;  and,  secondly,  that  we  are  dealing  not 
with  a  perfectly  fresh  surface,  but  with  one  a  few  seconds 
old,  and  during  this  initial  period  the  elasticity  is  increasing 
very  rapidly,  so  that  the  value  obtained  will  depend  upon  the 
exact  moment  of  taking  the  observation.  It  is  probable, 
too,  that  the   manner  of  formation  of  the  surface  has  some 
