I have to apologize. I also misinterpreted the graph. The confusing
graph is, in the Second Edition, Fig. 12-4 on p. 546. I've uploaded it
temporarily to http://eznec.com/misc/Kraus2_Fig_12-4.JPG.

The caption says that the gain is relative to a half wave dipole in free
space with the same power input. The numbers on the left side Y axis are
the numerical gain, 0 to 3, relative to a dipole. 0 represents a
numerical gain mulitplier of zero, or no field intensity at all. 1 is
the gain of a half wave dipole in free space (about 2.15 dBi). A value
of 2 represents a gain in field intensity by a factor of 2, or 6 dB
relative to a dipole. The right hand side Y axis labels are the gain in
dBi. Note that 2.1 dBi corresponds approximately with the value of 1.0
on the left side. The bottom horizontal line corresponds to zero field
strength -- a gain of minus infinity dBi -- *not* zero dBi as Richard
said, or zero dB relative to a dipole, which I initially assumed.

What I missed was that the gain is "in direction [phi] = 0", quoting
from the caption. So this isn't a graph of the maximum gain, but the
*gain in one specific direction* -- normal to the reflecting plane. At
0.5 wavelength spacing, the "gain in field intensity" (left set of Y
axis labels) is a *factor* of zero, meaning that the field strength is
zero, or minus infinity dBi. Sure enough, if you model the antenna, or
two elements spaced one wavelength, you find that the pattern has a null
directly broadside to the antenna ([phi] = 0). It has gain in other
directions, but that's not what the graph is showing.

Of course, any lossless antenna has a gain of 0 dBi in some directions.
In the case of the element and reflecting plane, the gain directly
broadside to the antenna has a gain of 0 dBi at spacings of roughly
0.425 and 0.575 wavelengths. There's no particular significance to this
-- the maximum gain is greater in other directions.

These gains and patterns can easily be seen with any modeling program,
including the EZNEC demo, by modeling a dipole over perfect ground. You
can also model two elements fed 180 degrees out of phase at twice the
spacing and no ground and see that the pattern is identical except for
being bidirectional.

Roy Lewallen, W7EL

Fred W4JLE wrote:
Roy, I looked at the graph and get a different interpretation. Every spacing
except 1/2 wave length spacing shows gain. That being the case the pattern
must be distorted for all cases except .5 wavelength.

I have Kraus 1950 edition

"Richard Harrison" wrote in message
...

Roy Lewallen
, W7EL wrote:
"I don`t have Kraus` 3rd edition (yet), but there`s graph on p 546 of
thye second edition which I suspect is the same as the one Richard is
referring to."

I`m sure that`s it. I have Kraus` 1950 edition of "Antennas" and the
identical groph is on page 327 in it.

If you look at the patterns of a 1/2-wavelength antenna at spacings of
1/4, 1/2, and 1/16 wavelengths spacing from a flat reflector nearby,
they are all nearly circular, indicating little distortion in their
unblocked direction.

Best regards, Richard Harrison, KB5WZI