We observed Jupiter on 17-19 July 1994 using NASA's Kuiper Airborne Observatory (KAO) deployed out of Melbourne, Australia. The KAO Echelle Grating Spectrograph (KEGS) uses an array detector with 128 spectral elements and 20 spatial elements. The spectral resolution was 0.19 cm(-1), equivalent to a resolving power of 6800, We detected three H2O emission lines near 7.7 mu m as well as numerous emission features of CH4. At the peak of the fallback phase for the ejecta from the G and K fragments, we infer temperatures between 1000 and 1200 K at the 3-mu bar pressure level on Jupiter. The H2O lines at 7.7 mu m were visible for only about 10 to 20 min. We obtained line-of-sight abundances of H2O for both G and K of 1.0 to 1.5 x 10(18) cm(-2). The inferred mass of H2O in each fragment ranges from 1.4 to 2.8 x 10(12) g, equivalent to spheres of ice 140 to 180 m in diameter at unit density. This should be considered a lower limit to H2O as the 7.7-mu m data are sensitive only to the hottest component of water. The G and K impact sites exhibited remarkably similar spectra at the time of peak infrared emission about 14 min after impact. The inferred abundance of H2O at this time was comparable to that of CH4. We interpret the observed CH4 emission as due to heating of preexisting jovian CH4 in the upper stratosphere. Water in the impact sites can be explained by shock chemistry provided that the C/O ratio is less than 1. Our observations are consistent with a scenario in which oxygen-rich cometary material combines with jovian H-2 to produce H2O during the fallback phase. Combining our observations with spectra acquired from the Near Infrared Mapping Spectrometer on the Galileo spacecraft, we do not believe that the fragments reached the 6-bar level in order to excavate jovian water. (C) 1996 Academic Press, Inc.