Spectrophotometric observations (1.7-12.6 mum) of SN 1987A from the Kuiper Airborne Observatory are presented for five epochs at 60, 260, 415, 615, and 775 days after the explosion. A variety of emission lines is seen, including members of the hydrogen Humphreys, Pfund, Brackett, and Paschen series, fine-structure lines of metals (including [Ni II] 6.634 mum, [Ni I] 7.507 mum, [Ar II] 6.985 mum, and [Co II] 10.521 mum), and CO and SiO molecular bands. The temporal evolution of the seven strongest H lines follows case C recombination theory and yields large values Of tau(Halpha) at 260 and 415 days. A mass of approximately 2 x 10(-3) M. is derived for stable nickel, and the ratio of the [Ni I] 7.507 mum and [Ni II] 6.634 mum line intensities yields a high ionization fraction of 0.9 in the nickel zone. Dust condensation is clearly detected at 615 days for the first time in a Type II supernova. At no time is there a 9.7 mum emission feature characteristic of interstellar astronomical silicates in the spectra of SN 1987A, nor are the 6.2 or 7.7 mum emission features attributed to polycyclic aromatic hydrocarbons seen. These airbome data are combined with other airbome and ground-based measurements taken at (or near) the same time to form five composite spectra of SN 1987A with wavelength coverage from approximately 3200 angstrom to 100 mum. The IR continuum emission between approximately 2 and 100 mum is compared with a three-component model-(1) hot photospheric continuum, (2) free-free and free-bound H emission, and (3) dust continuum-with the best fit determined using a nonlinear chi2 method. The dust continuum component is well characterized by a single-temperature graybody emission spectrum, i.e., by the radiation from gray grains or dust in optically thick clumps. At early times (less than 400 days after core collapse), the dust emission tracks the bolometric luminosity at about the 2% level. By 615 days, the fraction of the total luminosity contributed by the IR dust continuum increases dramatically to 0.45, and then to 0.83 at 775 days. We suggest that this dichotomy in the temporal evolution of the dust emission arises from dust with different origins. Circumstellar dust present before the supernova and then heated by it may account for the early emission. Newly condensed dust in the ejecta accounts for the later emission. A lower limit to the dust mass at 775 days is approximately 10(-4) M., but much more dust could be present. Since the emission is well fitted by a graybody, no information on the dust composition can be directly discerned from our data.
