Am'  ja°ar'i8P84arm'}  Thermometric  Measurements.  47 
of  needles,  and  contains  variable  amounts  of  water.  Caffeine  auro- 
ehloride,  C8H10N4O2,  HAuCl4, 2H20,  forms  lustrous  gold-colored  plates. 
Caffeine  methiodide,  C8H10N4O2,  Mel,  H20,  is  formed  when  caffeine  is 
heated  for  some  hours  at  130°  with  an  excess  of  methyl  iodide  in 
sealed  tubes,  and  may  be  purified  by  washing  with  cold  alcohol  and 
crystallizing  from  water,  in  which  it]  is  moderately  soluble,  although 
but  sparingly  so  in  alcohol,  and  almost  insoluble  in  ether. — Jour. 
Chem.Soe.,  Sept.  1883;  Annalen  217,  p.  270. 
THERMOMETRIC  MEASUREMENTS. 
By  J.  M.  Crafts. 
The  author  remarks  at  the  outset  that  the  progress  made  in  the 
purification  and  preparation  of  chemical  substances  has  not  been 
accompanied  by  any  appreciable  improvement  in  thermometric  meas- 
urements. In  these  communications  the  author  gives  an  account  of  a 
series  of  experiments  on  the  commonly  employed  methods  of  fusion 
and  ebullition,  with  a  view  of  facilitating  the  construction  of  ther- 
mometers, of  examining  their  behavior,  and  of  rendering  the  method 
of  observation  precise. 
In  the  thermometers  from  the  best  sources  the  author  observed 
residual  errors  of  0*015 — 0*04  degree,  and  when  the  scale  is  divided 
into  tenths  of  a  degree,  in  ordinary  thermometers  differences  of  length 
of  0*1 — 0'5  degree  in  contiguous  sections  of  25  degrees.  As  these 
variations  rarely  compensate  one  another,  it  is  not  rare  to  find  thermo- 
meters which  require  corrections  amounting  to  several  fractions  of  a 
degree. 
In  determining  the  value  of  a  degree  from  the  points  0°  and  100°, 
it  is  most  important  to  follow  an  invariable  order  in  the  observation  of 
these  points.  After  the  point  100°  has  been  fixed,  the  zero  point  must 
be  determined  by  quickly  cooling  and  placing  the  thermometer  in 
pounded  ice,  or  preferably  snow,  which  has  remained  for  some  time 
in  contact  with  distilled  water.  But  even  after  adopting  all  the  neces- 
sary precautions,  the  value  of  a  degree  'may  vary  owing  to  the  dis- 
placement of  the  zero  point,  which  causes  a  change  in  various  propor- 
tions of  all  the  constants.  In  the  original  paper  examples  are  given 
to  show  that  the  elevation  of  zero  of  1'24  to  2*6°  may  cause  an  increase 
in  the  interval  0—100°  of  0*04  to  0'9  degree.    On  the  other  hand,  a 
