562  PROF.  J.  W.  JUDD  ON  COMPOSITE  DYKES  IN  ARRAN.  [RoV.  I 893, 
complete  consolidation  and  crystallization  of  the  materials  of  the  one 
rock  must  have  occurred  before  its  invasion  by  the  other  rock  :  this 
is  proved  by  the  characters  of  the  junctions  and  also  by  the  derived 
crystals. 
"These  composite  dykes  afford,  moreover,  very  valuable  evidence 
concerning  the  relative  ages  of  some  of  the  larger  ejections  in  the 
area  of  the  West  of  Scotland.  I  have  shown  that  at  the  five  great 
centres  of  volcanic  outburst  in  that  area — at  Mull,  Ardnamurchan, 
Rum,  Skye,  and  St.  Hilda — the  order  of  appearance  of  the  three 
kinds  of  igneous  material  was  as  follows  : — intermediate,  acid,  basic. 
In  Arran,  however,  the  great  basic  eruptions,  represented  by 
olivine-gabbro  and  plateau-basalts,  seem  to  have  been  wanting. 
Farther  south,  in  Ireland,  both  acid  and  basic  rocks  are  represented 
by  granites  and  gabbros ;  but  whether  these  are  to  be  regarded  as 
contemporaneous  with  the  similar  rocks  in  the  Scottish  area  is  a 
question  which  has  been  left  unsettled  by  the  work  of  the  Geological 
Survey  in  that  region. 
That  the  rocks  with  which  we  are  concerned  in  this  paper  are  of 
far  more  recent  age  than  the  granite  of  Arran  is  evident,  because 
at  Cir  Mhor  we  "find  the  granite  intersected  by  this  and  similar 
dykes.  It  is  clear  that  not  only  was  this  great  mass  of  granite 
already  ejected,  but  that  it  must  have  become  consolidated  and 
acquired  its  present  structures.,  before  the  injection  into  it  of  the 
andesite  and 1 2 3  4  pitchston-e.’ 
Of  still  greater  interest,  as  I  have  already  intimated,  is  the  light 
thrown  by  the  study  of  these  dykes  upon  the  cause  of  the  differen¬ 
tiation  of  lavas — a  question  which  has  frequently  been  discussed  in 
the  past,  but  which  during  the  last  few  years  has  received  new 
treatment  at  the  hands  of  Guthrie,1  Lagorio,"  Teall,3  liosenbusch, 
Brogger,5  Vogt,6  and  Iddings.7 
We  may  classify  the  explanations  which  have  been  •  suggested  to 
account  for  the  differentiation  of  lavas  into  two  groups  : — those 
which  rely  upon  some  process  of  selective  crystallization,  and  those 
which  premise  some  kind  of  separation  taking  place  in  a  liquid  magma 
prior  to  the  commencement  of  crystallization  and  consolidation. 
Differentiation  during  crystallization  has  been  referred  to  two 
distinct  causes.  Prof.  Vogt,  in  order  to  account  for  the  deposits 
so  rich  in  magnetite  occurring  in  Scandinavia  and  elsewhere,  has 
put  forward  the  theory  that  the  remarkable  magnetic  properties  of 
the  iron-spinellid  may  account  for  its  segregation,  and  that  the 
mutual  attraction  of  the  magnetite-particles  may  lead  to  their 
1  Phil.  Mag.  5th  ser.  vol.  xvii.  (1884)  pp.  462-482,  vol.  xviii.  (1884)  pp.  22-35 
&  105-120. 
2  Min.  u.  petr.  Mitth.  vol.  viii.  (1887)  pp.  421-51.9. 
3  *  British  Petrography,’  1888,  pp.  391-403 ;  and  ‘  Natural  Science,’  vol.  l. 
(1892)  pp.  288-299. 
4  Min.  u.  petr.  Mitth.  vol.  xi.  (1889)  pp.  144-176. 
5  Zeitschr.  fur  Krvstallogr.  u.  Min.  vol.  xvi.  (1890). 
6  Geol.  Forening.  Forhandl.  vol.  xiii.  (1891)  p.  476. 
7  Bull.  Phil.  Soc.  Washington,  vol.  xi.  (1890)  pp.  191-220,  vol.  xii.  (1891) 
pp.  89-214. 
