NO.    I  ARCHEOLOGY    OF    ST.    LAWRENCE    ISLAND — COLLINS  257 
should  be  observed  in  the  theoretical  discussions  involving  Alaskan 
Eskimo  houses  is  perhaps  natural,  for  with  the  exception  of  the  Point 
Barrow  and  Mackenzie  types,  there  are  few  adequate  descriptions 
of  the  houses  in  this  region. 
In  1927,  while  making  a  reconnaissance  of  the  Alaskan  coast  from 
Bristol  Bay  to  the  mouth  of  the  Yukon,  I  was  able  to  obtain  detailed 
information,  including  photographs  and  measurements,  on  the  houses 
in  this  area,  where  dwell  the  most  primitive  group  of  Eskimos  remain- 
ing in  Alaska.  This  is  particularly  true  of  Nunivak  Island  and  the 
adjacent  mainland  ;  in  this  isolated  region  the  underground  houses, 
like  many  other  features  of  native  culture,  have  been  unafifected  by 
contact  with  civilization.  In  the  two  succeeding  years  I  examined  the 
coast  to  the  northward  and  obtained  information  on  house  types  on 
St.  Lawrence  Island,  at  Bering  Strait,  and  at  Point  Hope.  It  is  not 
my  intention  to  enter  here  into  a  detailed  discussion  of  the  Eskimo 
houses  of  southwest  Alaska,  but  since  it  is  essential  that  their  rela- 
tionship to  those  of  northern  Alaska  be  understood,  a  brief  summary 
and  comparison  will  be  required. 
It  is  essential  first  of  all  to  recognize  that  the  Eskimo  houses  south 
of  Bermg  Strait  dififer  fundamentally  from  those  to  the  northward. 
The  failure  to  recognize  this  distinction  on  the  part  of  most  of  those 
who  have  discussed  the  problem  of  Eskimo  houses  seems  to  be  due 
mainly  to  their  reliance  on  the  very  general  and  sometimes  misleading 
statements  of  Petroff,  rather  than  on  the  more  explicit  descriptions 
given  by  Sagoskin,  Nelson,  and  Gordon. 
Another  important  fact,  which  has  not  been  sufficiently  noted  here- 
tofore, is  that  the  ordinary  dwellings  in  this  region  dififer  in  certain 
specific  respects  from  the  ceremonial  houses  or  "  kashims."  Thalbitzer 
has  made  a  careful  study  of  the  ceremonial  house  among  the  Eskimo, 
tracing  its  distribution  from  the  fully  developed  stage  in  Alaska  to 
the  more  attenuated  forms  found  in  the  Central  regions  and  Greenland 
(1925,  pp.  236-255).  Architecturally,  these  northern  forms  of  the 
ceremonial  house  show  little  interrelation,  the  connection  being  pri- 
marily one  of  function.  There  is,  however,  a  rather  striking  agreement 
in  architectural  detail  as  well  as  in  function  between  the  kashims  of 
the  Briston  Bay- Yukon  region  of  Alaska  and  the  kivas  or  ceremonial 
rooms  of  the  Pueblo  Indians  of  the  Southwest.  This  agreement  con- 
sists of  a  high  cribbed  or  vaulted  roof,  a  central  fire  pit,  a  ventilator 
shaft  extending  from  the  fire  pit  to  the  outside,  and  a  low  platform 
or  bench  built  along  the  walls.  Except  for  the  platform,  the  other 
features  are  common  to  the  ceremonial  structures  in  both  regions  but 
are  lacking  in  the  dwellings.   I  will  not  discuss  here  the  possible  sig- 
