542  PROF.  J.  W.  JUDD  ON  COMPOSITE  DYKES  IN  ARRAN.  [HoV.  1 89 3, 
have  to  show  in  these  pages,  however,  what  an  important  part  hyalite 
and  tridymite  play  in  some  of  these  rocks. 
The  crystallized  minerals  which  are  present  consist  of  pyroxene 
(either  augite  or  enstatite),  magnetite,  titanoferrite,  quartz,  and 
numerous  felspars.  According  to  the  usually  accepted  systems  of 
petrographical  nomenclature,  the  name  to  be  given  to  these  rocks 
would  he  determined  by  the  presence  or  absence  of  free  quartz  and 
by  the  characters  of  the  felspar-crystals  ;  and  it  is  to  the  exact  deter¬ 
mination  of  the  felspars,  therefore,  that  especial  attention  must  be 
directed  in  hxing  the  place  of  these  rocks  in  the  petrographic  series. 
At  first  sight  it  appears  that  the  majority  of  the  felspar-crystals, 
which  are  scattered  through  the  glass,  are  destitute  of  lamellar 
twinning,  and  should  be  referred  to  orthoclase.  It  is  scarcely 
necessary,  however,  to  point  out  that,  while  the  presence  of  lamellar 
twinning  may  be  regarded  as  characteristic  of  plagioclase-felspar, 
the  absence  of  that  character,  as  Hawes  so  well  showed,  is  no  proof 
that  the  crystals  do  not  belong  to  the  triclinic  system.  A  very 
careful  examination  of  these  crystals  in  polarized  light,  with  properly 
adjusted  illumination,  shows  that  many  of  the  apparently  untwinned 
crystals  exhibit  the  delicate  lamellar  structure  so  characteristic  of 
‘anorthoclase’  or  ‘  cryptoperthite/  In  addition,  moreover,  to 
these  orthoclase-  or  anorthoclase- crystals,  a  considerable  number  of 
very  distinctly  twinned  plagioclase-crystals  nearly  always  occur, 
and  these  are  sometimes  so  abundant  as  to  predominate  over  the 
untwinned  felspar-crystals.  It  will  thus  be  seen  that,  taking  the 
crystallized  minerals  as  our  guide,  as  is  customary  with  petro- 
graphers,  and  remembering  that  in  some  cases  quartz- crystals  are 
present,  and  in  other  cases  entirely  absent,  the  rocks  would 
be  variously  classed  as  trachytes  or  rhyolites,  pantellerites .  or 
quartz-pantellerites,  andesites  or  quartz-andesites  (dacites),  according 
to  the  absence  or  presence  of  quartz,  and  the  predominance  of  ortho¬ 
clase,  anorthoclase,  or  plagioclase  among  the  felspars. 
That  these  rocks  are  not  simple  rhyolites,  as  has  sometimes  been 
assumed,  is  proved  by  the  fact  that  most  of  the  analyses  which  have 
been  made  of  them  reveal  the  presence  of  a  very  large  proportion  of 
soda,  and  in  many  cases  the  soda  predominates  over  the  potash. 
The  researches  of  Lagorio  show  that  the  potash  will  probably  abound 
in  the  glass,  and  the  soda  in  the  porphyritic  constituents.  Isolation 
and  analysis  of  the  felspars  in  these  rocks  are  usually  rendered 
difficult  by  the  frequency  in  them  of  glass  and  other  inclusions  : 
but  tests  by  Szabo’s  method  of  flame-reactions,  made  upon  examples 
selected  as  freest  from  inclusions,  show  that  these  felspars  are 
almost  always  very  rich  in  soda.  I  am  convinced  that  in  mam  case:? 
the  rocks  should  be  classed  with  the  ‘  pantellerites  ’  and  ‘  quartz- 
pantellerites/’  and  this  view  is  confirmed  by  the  most  trustworthy 
of  the  analyses  which  have  been  made  of  them. 
But  no  one  can  study  these  rocks  without  being  convinced  that 
it  is  the  glassy  base ,  and  not  the  crystallized  minerals  scattered 
through  it,  which  constitutes  their  distinctive  feature  ;  and  that  it  is 
this  groundmass,  and  not  the  enclosed  crystals,  which  ought  to  be 
