The moons of Mars are not quite like our Earth's Moon. Phobos, the larger of the two, is much closer to its planet; compared to the Moon's 27-day orbit, Phobos swings around Mars in line with the planet's equator thrice every Martian day (sol).
Solar eclipses, therefore, are much more frequent than those here on Earth. Phobos passes in front of - but never entirely covers - the Sun for an annular or partial eclipse somewhere on Mars most sols. Because Phobos is moving so fast, it never transits for more than 30 seconds.
But, even during this brief time, the Mars InSight lander has recorded something peculiar happening.
To the surprise of Mars scientists, during Phobos eclipses, the lander's seismometer - the instrument that records ground motions to monitor possible quake activity - tilts, just an infinitesimal little bit, towards one side.
Researchers at ETH Zurich's Institute of Geophysics were actually studying data from Mars InSight to see if some of the effects of eclipses here on Earth also occur on Mars.
Specifically: "When Earth experiences a solar eclipse, instruments can detect a decline in temperature and rapid gusts of wind, as the atmosphere cools in one particular place and air rushes away from that spot," explained seismologist Simon Stähler of ETH Zurich.
InSight is equipped with temperature and wind sensors - but these recorded no change in the atmosphere during Phobos transits. Atmospheric turbulence, atmospheric temperature, and barometric pressure remained pretty much consistent with a normal sol.
The solar cells did, however, register the transits. Actually, it would be very curious if they didn't, since Phobos can block as much as 40 percent of the Sun's light - so it was reassuring that something went according to plan.
"When Phobos is in front of the Sun, less sunlight reaches the solar cells, and these in turn produce less electricity," Stähler said. "The decline in light exposure caused by Phobos's shadow can be measured."
But that was the extent of the "expected". Because both the magnetometer and the seismometer chimed in with odd readings - the seismometer with its unexpected tilt.
Actually, the oddity with the magnetometer - used to monitor the magnetic field on the Martian surface - was pretty easy to figure out. Two components showed a decrease very similar to the decrease in the current from the solar array. So the scientists deduced that the decreased current was likely the cause.
"But we didn't expect this seismometer reading; it's an unusual signal," Stähler said. "Imagine a 5-franc coin; now, push two silver atoms under one edge. That's the incline we're talking about: 10^-8."
It doesn't seem to be a false positive; the signal is recorded for three transits, faint but real. The team expected that it might be a seismic response to the moon's tidal - that is, gravitational - pull as it passed overhead.
However, when they compared it to other readings of seismic activity from Mars, the signal bore no similarity to previous seismic activity.
Another possibility is that the tether connecting the seismometer to the lander contracted. However, this would have produced a tilt in the opposite direction to what was observed.
And a change in atmospheric temperature could have introduced a density change that nudged the seismometer, but, as we have already discussed, no such change was detected.
But there was one more signal. An infrared radiometer recorded a slight drop in surface temperature during the longest transit, followed by a period of about a minute and a half while the ground warmed back up to its pre-transit temperature.
This, the team believes, is the most likely cause of the strange reading.
"During an eclipse, the ground cools," said seismologist Martin van Driel of ETH Zurich. "It deforms unevenly, which tilts the instrument."
A similar effect was observed in 1997, at the Black Forest Observatory in Germany.
A technician forgot to turn off the light when leaving the seismometer vault, resulting in elevated noise in long-period data as the warmth from the bulb expanded the granite on which the seismometer rested.
A series of experiments with artificial heat sources ensued, demonstrating that seismometers react almost instantaneously to heat changes in the seismic pillar.
The team repeated their own version of this experiment, and found that they were able to obtain a signal consistent with the tilting of InSight's seismometer.
This information could be used to better understand Phobos and Mars, the researchers said.
For one, InSight's location is very accurately mapped. Knowing when a Phobos eclipse begins and ends at that location could help scientists more accurately constrain its orbit.
And that, in turn, could help us understand what's in store for Phobos' future.
The moon's orbit is decaying at a rate of 1.8 centimetres per year, slowing as it goes; eventually, scientists predict, it will grow so close to Mars that tidal forces will tear Phobos apart, turning it into a ring of debris circling Mars.
If the slowdown can be characterised, that can tell us how elastic and warm the planet's interior is - or how inelastic and cool. And that, in turn, can shed some light on Mars' formation history.
The research has been published in Geophysical Research Letters.