The modulation lanes in Jupiter's decametric radio spectra were discovered by Riihimaa [1968>. We have developed a model for the mechanism responsible for their production in which the free parameters have been adjusted to provide a very close fit with the observations. In our model, a grid-like interference screen composed of field-aligned columns of enhanced or depleted plasma density is located near Io's orbit close to the longitude of the sub-Earth point. The column spacing is typically about 140 km. As a band of frequency components emitted from near the foot of an excited tube of magnetic flux passes through the screen, interference patterns of slightly different orientations are produced by the different frequencies. The corotation of this set of interference patterns with Jupiter results in the sloping modulation lanes of the observed dynamic spectrum. Newly calculated results indicate that (1) the Io-B and Io-A radiations are emitted from the northern hemisphere, while that from Io-C comes mainly from the southern hemisphere, (2) the half-angle of the assumed hollow-cone emission beam for Io-B is typically 60¿, with a variation of a few degrees, and (3) the equatorial lead angle of the radio-emitting previously excited flux tube ahead of the flux tube through Io at the same instant is more variable, 50¿ being a typical value for Io-B. Modulation lanes from Io-unrelated emission were also successfully modeled, the assumption in this case being that some source of energy other than a direct encounter with Io excited the flux tube containing the emitting radio source. The L shell value for non-Io-A was less clearly defined but was definitely between 4 and 7. Sensitive measurements of the lane modulation depths for Io-B and Io-A indicated that the decay time for the columns of the interference screen is of the same order of magnitude as Jupiter's rotation period. Although the predominant modulation lane periodicity of about 2 s indicates an upper limit on Io-B source width of 70 km along the direction of increasing longitude, finer structure that was present in at least one case suggests an upper limit of only 20 km.¿ 1997 American Geophysical Union