Analysis of the data returned from the three Rice lunar surface Suprathermal Ion Detector Experiments (Side) has shown that the boundary layer and plasma mantle observed near the earth extend to the lunar orbit as an extensive region of antisunward plasma flow occupying a major portion of the lobes of the geomagnetic tail. At lunar distances we call this the lobe plasma. The lobe plasma appears as a narrow peak in the ion energy spectrum usually around 100 eV. More recent analysis of Side data has revealed the occasional occurrence of a second peak in the ion spectrum during geomagnetically active periods. This peak is similar to a secondary peak reported by Frank et al.  in the boundary layer energy spectrum. This secondary peak occurs at higher energy than the principal peak, generally in the range 1–2.75 keV. Both spectra are narrow, the upper peak frequently being seen in one or two channels of the analyzer. The inferred temperature for both peaks is usually less than kTi ∼ 10 eV in the look direction of the instrument. The differential ion flux for the secondary peak is estimated to be generally less than 3% of that for the primary peak. On the assumption that both peaks arise from flowing protons we find that the ratio of the bulk velocity of the secondary peak to that of the primary peak ranges between 4.4 and 5.4. We suggest two hypotheses for the origin of these ions. Either (1) they consist of two streams of protons traveling with different bulk velocities and therefore originating from two different source regions, or (2) they are two streams traveling with approximately the same bulk velocity but with the primary peak composed of protons and the secondary peak composed of more massive ions. By using the second hypothesis (same bulk velocity) the Side data can be fitted well if a lunar surface potential between +35 and +91 V is assumed and if the secondary peak ions have a mass of 14 or 16 and unit charge. The most likely candidate is ionospheric O+. On the assumption that the the secondary peak is O+, by fitting to a convected Maxwellian we obtain ion number densities between 5 × 10−5 and 10−3/cm³. These numbers should be treated with caution because of the narrowness of the ion angular and energy distribution. In both hypotheses an ionospheric source is postulated, and an energization process is needed to account for the observed ions.