O-C diagram of the principal superhump of AM CVn in 1998. Each
point is a nightly pulse timing, with an average length of about 3.6
hr. The test period is 525.55 s, so the very slight downward trend
indicates a mean period very slightly less. This is in sharp contrast
to 1997, when the superhump occurred at the highest period (525.65 s)
seen in any season of dense observation.
The point-to-point scatter is ±26 s. Measurement error is estimated
at ~15-20 s. So the extra error incurred by using nightly runs (rather
than full 13.38 hr runs which fully resolve any sideband structure) is
around 15 s -- essentially buried in the noise. This is why pulse
period changes in this star are so easy to follow!
We'll continue the campaign through early July. This is a great, great
target for regular monitoring with small telescopes, because the
shortness of the period makes it pretty easy to get a decent pulse
timing with just a few hours per night.
O-C diagram of the 1011 s ("negative") superhump of AM CVn in
1998. The test period is 1011.40 s, so the probable decline
signifies a slightly shorter period. The timing errors are
about 3 times greater here, because: the signal is weaker;
the period is longer (fewer cycles to average over); and
there is a neighboring signal at 1028 s which is hard to
The general drift in the period (given by the amplitude and
timescale of the O-C wiggles) is fairly similar to that of
the 525 s superhump (which is much more richly documented
since it does a basically similar dance every year). But we
want to explore in detail how the two sets of wiggles
correlate. To succeed in this we need to observe the star
very frequently. As you can see a little wiggle occurred
near E = 2000 which is not defined with certainty by our data --
there is some chance that the points belong exactly one
cycle lower, or even half a cycle (with sudden transitions).
We want to eliminate (or confirm!) such possibilities to get
a really definitive test of the correlation between the two
O-Cs. It looks to me like diligent observation for the rest
of the season will preserve cycle count and hence define the
The perpetrators of this most promising work have been:
Jonathan Kemp, Dave Skillman, Dave Harvey, Bob Fried, Tonny
Vanmunster, and Gianluca Masi. We'd love to get more help!
Here are the three most interesting regions of the power spectrum
during JD 945-952. These are essentially "cleaned" power spectra,
i.e. with the aliases of each detected signal removed. The upper frame
shows all three primary signals: the orbital signal at 1028.7 s, the
positive superhump at 1051.2 s, and the negative superhump at 1011.4 s.
These are the familiar basic signals, although we never previously saw
them manifest at the same time. In our terminology, we call w the
orbital frequency, W the frequency of apsidal advance, and N the
frequency of nodal regression. The superhump frequencies then
correspond to w-W and w+N.
The middle frame shows the vicinity of the strongest signal at 525.6 s
(the first harmonic of the positive superhump), which rises to a power
of 255. There are signals at 520.0 and 515.4 s (166.17 and 167.63
c/d). The former is the 2w-W signal, occasionally seen in superhumping
CVs. But the latter is certainly surprising. Probably it is 2w-W+N.
The third frame shows the second harmonic of the positive superhump, at
350.4 s, and a weak extra signal at 345.3 s (250.19 c/d). The latter
The semi-amplitude of the 525.6 s signal is 0.012 mag. Since power
scales with (amplitude)2,
this means that the 1028 and 1051 s signals
have a semi-amplitude of 0.0046 mag, and the weakest signal flagged
has a semi-amplitude of 0.0018 mag.
The phase of the orbital signal was consistent with the ephemeris of
Harvey et al. (1998). The phase of the other signals wandered slowly
throughout the season, as usual.