Radial velocity measurements of stellar reflex motion have revealed many extrasolar planets , but gaps in the observations produce aliases , spurious frequencies that are frequently confused with the planets ’ orbital frequencies .
In the case of Gl 581 d , the distinction between an alias and the true frequency was the distinction between a frozen , dead planet and a planet possibly hospitable to life ( ) .
To improve the characterization of planetary systems , we describe how aliases originate and present a new approach for distinguishing between orbital frequencies and their aliases .
Our approach harnesses features in the spectral window function to compare the amplitude and phase of predicted aliases with peaks present in the data .
We apply it to confirm prior alias distinctions for the planets GJ 876 d and HD 75898 b .
We find that the true periods of Gl 581 d and HD 73526 b/c remain ambiguous .
We revise the periods of HD 156668 b and 55 Cnc e , which were afflicted by daily aliases .
For HD 156668 b , the correct period is 1.2699 days and minimum mass is ( 3.1 \pm 0.4 ) M _ { \oplus } .
For 55 Cnc e , the correct period is 0.7365 days – the shortest of any known planet – and minimum mass is ( 8.3 \pm 0.3 ) M _ { \oplus } .
This revision produces a significantly improved 5-planet Keplerian fit for 55 Cnc , and a self-consistent dynamical fit describes the data just as well .
As radial velocity techniques push to ever-smaller planets , often found in systems of multiple planets , distinguishing true periods from aliases will become increasingly important .