DOC.
153
MAY 1909
111
a
+
b
Fig. 1.
in the
retina,
I would have
expected
these numbers
to
be
greater.
However,
one
cannot
attach
great
importance
to this.
As
regards
the
spreading
in
the direction of
propagation,
let
me
point out
that
interference
phenomena up
to
a
phase
difference of
2 million
wavelengths
have
been
observed in
"homogeneous"
light.
From
this
one can
conclude
that the
rays
employed
must
display
at
least
2
million
waves
(and probably
even
somewhat
more)
following
regularly
upon
each
other,
and
this conclusion holds for
each
individual
light
quantum,
because if
one
quantum
would not
manifest
any
noticeable
interference,
the combined
action
of
several
quanta
would not be able to
bring
it
about either. For
we
must conceive
of the individual
quanta
as being totally
independent
of each other and hence
not
"interferable." If
we
take the
wavelength
to
be 0.4
fi,
then 2.106
waves
will
make for
a
stretch of
80
cm.
In order
to
understand that
a
light
quantum must
also extend
quite
far
in
directions
perpendicular
to
the
rays,
let
us
consider
a
telescope
with
an
objective
radius of
50
cm.
Let the
telescope
be
so
"good"
that
the full
opening
of the
objective
is
useful,
i.e.,
that
the
image
of
a
star
is
as good as can
be
expected according
to
the
customary, empirically
confirmed,
diffraction
theory.
Experience
shows
that
if
one
places
a screen
with
an
opening
like ab in
front of
the
objective,
the
image
becomes
definitively
poorer
than
with
the
unscreened
objective. If,
in
the latter
case,
the
light
is
restricted
to
the
small
part pq
of
the
focal
plane,
it
spreads
over a
larger part
of the
surface,
say rs,
after
the
placing
of the
screen.
I assert
that
if the
various
light
quanta
incident
on
the
(unscreened)
objective
had
a
lateral
extension
no
greater
than
ab,
then
the
distribution
of the
light
in
the
focal
plane
would
have
to
have
the
extension
rs
and could
not
be
concentrated
upon
the smaller
surface
area
pq.
That
is to
say
that
a
number of
light
quanta
(with
the
same
lateral
extension)
such
as
ab,
a'b',
a"b",
etc.
(Fig. 2)
could
never
be
concentrated
all
together
in the
area
pq
if this
were
not to
happen
with
each of
the individual
quanta,
and if
the latter
were
the
case,
then
a
number of
quanta
that
fell
into the
screen
opening
ab in
Fig.
1
would also have to be
concentrated
in
pq.
From
this
analysis
I
con-
clude
that
for
a
telescope to provide
images
of
a quality
corresponding
to
its
opening,
all
or
almost
all
quanta
it receives must
cover
the
full
P q
Fig.
2
P
q