563 
Vol.  49.]  PROF.  J.  W.  JUDD  ON  COMPOSITE  DYKES  IN  ARRAN. 
accumulation  at  certain  points  in  the  still  fluid  magma.  But  this 
ingenious  suggestion  omits  a  very  important  consideration.  It  has 
been  shown  by  Prof.  Hiicker,  F.R.S.,  that  both  metallic  iron  and 
magnetite,  when  heated,  rapidly  lose  their  magnetic  properties, 
and  that  at  a  dull  red  heat  all  phenomena  both  of  attraction  and 
polarity  disappear  alike  in  metallic  iron  and  in  magnetite.  Such 
being  the  case,  it  seems  impossible  to  conceive  of  such  a  degree 
of  heating  of  a  magma  as  would  permit  of  internal  movements, 
and  yet  would  not  at  the  same  time  destroy  the  attractive 
properties  of  the  magnetite. 
The  other  suggested  method  by  which  differentiation  may  be 
brought  about  in  a  magma  during  the  process  of  crystallization,  is 
based  on  the  fact  that,  as  a  general  rule,  the  more  basic  minerals 
crystallize  out  of  a  molten  mass  before  those  of  acid  composition,  and 
that,  as  a  consequence  of  this,  the  matter  left  liquid  continually  tends 
to  become  more  and  more  acid  in  composition.  That  in  these  masses 
of  molten  rock  consisting  of  basic  crystals  diffused  through  an  acid 
menstruum,  a  process  of  partial  or  complete  liquation  may  occur,  is 
shown  by  the  phenomena  which  have  been  observed  at  Santorin,1 
Krakatoa,2  Beiun  Hiant,3  Garabal  Hill,4  etc.  This  method  of 
differentiation,  while  it  affords  a  satisfactory  mode  of  explanation  of 
the  first  class  of  composite  dykes — that  so  well  described  by  Lawson 
and  Vogt — is  clearly  inapplicable  to  the  cases  in  Arran,  where  the 
porphyritic  crystals  and  glassy  basis  in  the  two  constituent  rocks 
are  so  strikingly  contrasted  one  with  another. 
AVe  are  thus  compelled  to  fall  back  upon  the  view  that  an  actual 
separation  takes  place  among  the  materials  of  a  molten  magma 
before  the  work  of  crystallization  has  commenced ;  and  I  may,  in 
conclusion,  pass  in  review  the  several  suggestions  that,  have 
been  offered  to  account  for  such  action  in  a  mixed  mass  of  molten 
silicates. 
Bunsen  and  many  authors  since  his  time  have  dwelt  upon  the 
fact  that  such  molten  masses  of  silicates  are  really  solutions,  and 
must  obey  all  the  laws  which  govern  solvent  action.  Guthrie  and 
Lagorio  have  especially  insisted  on  this  view  of  the  subject. 
Many  writers  on  the  question  have  been  impressed  by  the  view 
that  the  two  magmas  into  which  a  mass  of  molten  silicates  may  be 
supposed  to  break  up  must  vary  greatly  in  density  ;  and  several  of 
these  writers  have  suggested,  in  more  or  less  distinct  terms,  that  this 
difference  in  specific  gravity  may  be  the  efficient  cause  of  their  sepa¬ 
ration.  But  until  recently  it  was  difficult  to  realize  the  physical 
principles  that  would  cause  density-differences  to  come  into  play  as 
factors  of  differentiation.  The  researches  of  Gouy  and  Chaperon,5 
however,  supply  a  possible  explanation  of  the  phenomenon.  These 
authors  have  pointed  out  that,  in  accordance  with  well-recognized 
1  ‘  Santorin  et  ses  Eruptions/  Fouque,  1879. 
2  Geol.  Mag.  for  1888,  p.  1. 
3  Quart.  Journ.  Geol.  Soc.  vol.  xlvi.  (1890;  pp.  371-381. 
4  Idl'd,  vol.  xlviii.  (1892)  p.  104. 
6  Ann.  de  Chiinie  et  de  Physique,  6th  ser.  vol.  xii.  (1887)  pp.  384-394. 
