Those voltage changes are excessive. Furthermore, those
voltages may have been larger than 9 volts, but for earth
ground which is keeping those voltage differences from
becoming excessive. IOW it may be a human safety problem.
gfretwell has provided an accurate solution that should not be
ignored. You have numbers and can reproduce the problem. The
utility therefore would have no problem finding and fixing the
problem the first time. Symptoms that are unacceptable and
should not be ignored.

This post assumes those numbers were recorded at the GE
panel.

wrote:
Short story: On a US-standard 240/120 100 A service, I am seeing 8 to 9
volts difference between phases when a 120 V 1.3 kW microwave is
running. Is this excessive, or acceptable?

Long story:

For the past few mornings, the UPS (APC Smart-UPS 600) in my office has
been switching on to battery power for a few minutes, even when the
lights in the rest of the house hadn't flickered or gone out. I
finally put a meter (Metex ME-11 DMM) on the wall outlet supplying the
UPS while the UPS was running and found out it was probably tripping
on _over_voltage: the wall socket was delivering over 130 volts!

After a bit of experimentation, I discovered that the voltage went up
when the microwave in the kitchen was running, which is why it happened
in the mornings when the other half was making tea. The microwave is on
one of the 20 A kitchen circuits, while my office is on a 20 A
receptacle circuit with a couple of other rooms. It's a GE microwave,
about 10 years old, 120 V 1300 W per the nameplate. I also noticed that
the voltage rise was less pronounced when the central air conditioner was
running. The compressor is 240 V and is fused and breakered at 40 A.

Suspecting a loose neutral, I inspected the breaker panel. Nothing
looked out of place. I tried tightening all of the screws on the
neutral/ground bus bar. They all took a bit of tightening - less than
1/8 turn - but none were really badly loose. I made very sure that the
neutral coming in from the meter was tight at the bus bar. I have the
US-standard 240/120 service at, I believe, 100 amps, and a GE split-bus
panel. It is original to the house (1969), although the breakers have
all been replaced. To my knowledge, the kitchen circuits are two
separate 20 A circuits with their own NM cable, not split like a
Canadian kitchen circuit.

After I tightened the screws, I went around in the house and made sure
most things were turned off. There were still a few computers running,
but the TV, stereo, lights, etc were all off. The air conditioner was
also shut off. I measured the voltage between the neutral/ground bus
bar and each hot lug with the microwave off and then with the microwave
running, heating about 8 oz (230 mL) of water in a coffee cup. I then
turned the A/C on and tried again. Results...

A/C uWave left right l-r notes
off off 122.5 123.2 -0.7 run 1
off on 126.3 117.6 8.7

off off 121.9 121.3 0.6 run 2
off on 126.1 116.7 9.4

on off 119.6 123.3 -3.7
on on 123.7 118.5 5.2
...