DOC.
393
OCTOBER
1917
389
t44
=
0
for
guv
and not
the
expression you
obtain.[3]
Since
I
do not know
where
the
inconsistency lies,
I
would
like
to
present my
calculations to
you quite
com-
prehensively.
Formula
(50)
initially produces
to
the
approximation
indicated
by
you
Kt.
ßß
pa
-E ßß
p4
Z
*ßP
a
ß
ßß
4 ß
=
-X"(dg^a
+
8gßa
-
dg^\
(dgȧ
.
d9aß
_
dg»a\
_
8
aßtx
V
9xß
dxß
dx*
)
V
dx»
d'Xß
dxß
)
v-'
(%4
.
dgßi\
Í^9ß4
\
iß
V
dxß dx»
)
\
dx»
dxß /
The last
sum
is
equal
to
zero,
as can
be
seen
by exchanging u
with
ß.
By
reformulating
the
first
sum,
we
find
ewgk[gp
wegokweg
w[egkiweg w[egtweg p
~t{~~_(8g~~~
c43~i 8x13
8Xc,~
~
f~
(1)
For
a
point
on
the
x1-axis, expressions
(70)
of
your
Annalen
paper[4]
yield
dgi2
^
dgn
=
dgn
_
dgu
_
a_
dx2
dxz
r2
’
dx\
r2 ’
dxi
r2
(2)
All
the other
dguv/dxa’s
are zero
at such
a
point.
For
a
set of
values, a, ß, u,
inserted after
the
summation
sign
in
(1)
to
con-
tribute
to
kt44
a
component differing
from
zero,
at least
one
of
the
indices
a,
B,
u,
must be
equal
to
1;
if all
three
are
not
equal
to
1,
the
remaining
two
must
equal
each other. Hence
the
following components
to
kt44
result: