As the other folks have said, you are receiving an image response when
you are tuned 910kHz below the actual frequency. This radio is using
high-side injection, which means that the LO is operating 455kHz above
your received frequency. If you look at the response of the mixer stage
of your receiver with a spectrum analyzer, you will see three dominant
signals; these will be the upper sideband signal (the image), a
suppressed LO signal, and the lower sideband signal (the desired
signal). Since the LO is operating 455kHz ABOVE the desired signal, and
the upper sideband response is 455kHz above the LO signal, it becomes
clear about that 910kHz image response.
There are a couple of reasons that double conversion is used; these are
equally important.
The first reason is that of image rejection. If a 1st I.F. of 45MHz
were to be used, the image response would be 90MHz above your desired
frequency. A low-pass filter with a 33MHz cutoff frequency ahead of the
front end would reject that image, because the response would be so far
down on the filter slope that you wouldn't know it was there, unless you
injected a very strong signal into the front end of your receiver.
The second reason to use a double conversion system is that of gain
distribution...........you generally don't want to have more than 70dB
of gain at any one frequency in a receiving system. By distributing the
gain between two different frequencies, an extra measure of stability
can be obtained. Now, about that 910kHz image............with a double
conversion system, you could still have that image if there isn't enough
selectivity at that 1st I.F. Nowadays this is a non-issue, because with
the use of monolithic crystal filters at the 1st I.F. the delta 910kHz
image response is pretty significant. If you look at crystal filter
specs for roofing filters that are intended to be used in a double
conversion system with a 455kHz 2nd I.F. the manufacturers will usually
give a suppression spec at that 910kHz offset. 50 to 60dB is a
reasonable spec, although the 8-pole crystal filters can have an 80dB spec.
Some of the more expensive receivers, such as those made by companies
such as Racal, Harris, Norlin, Watkins-Johnson, Rockwell-Collins, Cubic,
and others will use these multi-pole filters at the 1st I.F.
About that 9990kHz response of the 10MHz signal..........since you are
using a long wire antenna, fundamental overload could be the problem,
I.F. filter feedaround could be another, or as Mr Doug Smith stated, it
could be an image response. You would be amazed at how a receiver falls
apart (figuratively speaking), when you inject a very high level signal
into the front end.
I hope this helps!

Pete