We present results of high time resolution photometry of the eclipsing pre-cataclysmic variable NN Ser .
NN Ser is a white dwarf/M dwarf binary with a very low-mass secondary star ( \sim 0.2 { M _ { \odot } } ) .
We observed 13 primary eclipses of NN Ser using the high-speed CCD camera ULTRACAM and derived times of mid-eclipse , from fitting of light curve models , with uncertainties as low as 0.06 s. The data show that the period of the binary is decreasing , with an average rate of \dot { P } = ( 9.06 \pm 0.06 ) \times 10 ^ { -12 } s / s , which has increased to a rate of \dot { P } = ( 2.85 \pm 0.15 ) \times 10 ^ { -11 } s / s over the last 2 years .
These rates of period change appear difficult to reconcile with any models of orbital period change .
If the observed period change reflects an angular momentum loss , the average loss rate ( \dot { J } = 1.4 \pm 0.6 \times 10 ^ { 35 } ergs ) is consistent with the loss rates ( via magnetic stellar wind braking ) used in standard models of close binary evolution , which were derived from observations of much more massive cool stars .
Observations of low-mass stars such as NN Ser ’ s secondary predict rates of \sim 100 times lower than we observe .
The alternatives are either magnetic activity-driven changes in the quadrupole moment of the secondary star ( ) or a light travel time effect caused by the presence of a third body in a long ( \sim decades ) orbit around the binary .
We show that Applegate ’ s mechanism fails by an order of magnitude on energetic grounds , but that the presence of a third body with mass 0.0043 M _ { \odot } < M _ { 3 } < 0.18 M _ { \odot } and orbital period 30 < P _ { 3 } < 285 years could account for the observed changes in the timings of NN Ser ’ s mid-eclipses .
We conclude that we have either observed a genuine angular momentum loss for NN Ser , in which case our observations pose serious difficulties for the theory of close binary evolution , or we have detected a previously unseen low-mass companion to the binary .