The BL Lac-type active galaxy OJ 287 exhibits a 12 year periodicity with a
double-peaked maxima in its optical flux variations. Several models sought
to explain this periodicity, the first one firmly established in any active
galactic nucleus (AGN), as a result of the orbital motion of a pair of sup
ermassive black holes. In one class of models the orientation of the jets c
hanges in a regular manner, and the optical flaring is due to a consequent
increase in the Doppler boosting factor. In another class of models the opt
ical flaring reflects a true increase in luminosity, either due to an enhan
ced accretion during the pericenter passage or due to a collision between t
he secondary black hole and the accretion disk of the primary black hole. H
owever, these models have been based solely on the optical data. Here we co
nsider the full radio flux density monitoring data between 8 and 90 GHz fro
m the Michigan, Metsahovi, and Swedish-ESO Submillimeter Telescope AGN moni
toring programs. We find that the radio flux density and polarization data,
as well as the optical polarization data, indicate that the first of the t
wo optical peaks is a thermal flare occurring in the vicinity of the black
hole and the accretion disk, while the second one is a synchrotron flare or
iginating in a shocked region down the jet. None of the proposed binary bla
ck hole models for OJ 287 offers satisfactory explanations for these observ
ations. We suggest a new scenario, in which a secondary black hole penetrat
es the accretion disk of the primary during the pericenter passage, causing
a thermal flare visible only in the optical regime. The pericenter passage
enhances accretion into the primary black hole, leading to increased jet f
low and formation of shocks down the jet. These become visible as standard
radio and optical synchrotron flares roughly a year after the pericenter pa
ssage and are identified with the second optical peaks. In addition to expl
aining the radio and the optical data, our model eliminates the need for a
strong precession of the binary and for an ultramassive (greater than or eq
ual to 10(10) M.) primary black hole. If our interpretation is correct, the
next periodic optical flare, a thermal one, should occur around 2006 Septe
mber 25. Nonthermal, simultaneous optical and radio flares should follow ab
out a year later.