We used the database of eclipsing binaries detected by the OGLE microlensing project in the pencil-beam search volume toward Baade ’ s Window to define a sample of 74 detached , equal-mass , main-sequence binary stars with short orbital periods in the range 0.19 < P < 8 days .
The logarithmic slope of the period distribution , \log N \propto ( -0.8 \pm 0.2 ) \log P , was used to infer the angular-momentum-loss ( AML ) efficiency for the late , rapidly-rotating members of close binaries .
It is very likely that the main cause of the negative slope is a discovery selection bias that progressively increases with the orbital period length .
Assuming a power-law dependence for the correction for the bias \propto - C \log P ( with C \geq 0 ) the AML braking-efficiency exponent \alpha in dH / dt = P ^ { - \alpha } can take any value \alpha = -1.1 ( \pm 0.2 ) + C .
Very simple considerations of discovery biases suggest C \simeq 4 / 3 , which would give an AML braking law very close to the “ saturated ” one , with no dependence on the period .
However , except for plausibility arguments , we have no firm data to support this estimate of C , so that \alpha remains poorly constrained .
The results signal the utmost importance of the detection bias evaluation for variable star databases used in analyses similar to the one presented in this study .