Recent studies have suggested that the ambient interplanetary magnetic field (IMF) drapes about fast coronal mass ejections (CMEs) as they plow out through slower moving, quiescent solar wind. In this study we examine ISEE 3 plasma and magnetic field data in order to look for observational evidence of IMF draping ahead of fast CMEs. Since the upstream IMF is excluded from CMEs, the ambient field must drape about these ejecta and connect back to the Sun around their edges. For example, for a purely radial, inward directed IMF, this would mean taht a CME which extends into the ecliptic plane while being directed northward of its hould give rise to a southward Bz perturbation in the ecliptic plane ahead of the CME. While purely radial IMF configurations are rare, the radial component of the IMF should also give rise to such Bz perturbations.

In reality, it is difficult to make an unambiguous identification of a CME's source region, and the draped field sampled along a spacecraft trajectory ahead of a CME can be very complicated, including contributions from a wide range of heliolatitudes as well as heliolongitudes. In spite of these complications from a wide range of heliolatitudes as well as heliolongitudes. In spite of these complications we set out to tes the usefulness of the draping scenario for predicting the Bz perturbations (and hence associated geomagnetic activity) ahead of fast CMEs which produce interplanetary shocks using ISEE 3 plasma and magnetic field data and Cane's (1985) list of shocks associated with interplanetary type II radio bursts. Of Cane's 45 events, there were 26 events which were suitable for testing the draping hypothesis.

However, for only 17 of these 26 events was there a substantial change in the average Bz between the regions upstream from the shocks and between the shocks and CMEs. Of these 17 events the simpleminded radial component predictor developed in this study correctly predicts the direction of the Bz perturbations for 13 events (76%). While this result is certainly not conclusive, we consider it to be supportive of the draping scenario. In terms of using the draping model for real-time geomagnetic activity predictions, however, the outlook is not nearly so good since nine of the 26 events had no substantial change in the average Bz and since there is no a priori way to tell how substantial draping effects will be for a given case. ¿ 1989 American Geophysical Union