Most past work on infrared-luminous galaxies ignored the contribution from older stars in heating the interstellar dust. Past work on galaxies of all types has also usually adopted a physically unrealistic model for the internal dust distribution, which can lead to a misinterpretation of broad-band colors. We present and discuss U (0.36 μm), R (0.65 μm), J (1.25 μm), H (1.65 μm), and K (2.2 μm) images of the merging luminous galaxy NGC 6240, which allow us to study its stellar mass and light distribution, and to consider the usually neglected role of the older stellar population in heating the interstellar dust in merging systems. We estimate that older stars contribute between one-fifth and one-half of the total heating of the dust that is observed at far-infrared wavelengths in NGC 6240, while active star formation contributes roughly another fourth. Therefore, in contrast with almost every other recent study of this object, we argue that the system may not be undergoing a truly extraordinary burst of star formation but rather that much of the significant far-infrared emission is simply reradiated starlight from a badly disrupted and dispersed dusty interstellar medium. As part of our analysis, we develop a technique for estimating total stellar luminosities using visual and near-infrared photometry of systems whose star-forming histories and, hence, stellar populations are poorly known. Discovering that stars are forming within a colliding galaxy pair is insufficient evidence that the collision triggered the star birth. The close association that is commonly drawn in the literature between mergers/ interactions and consequent energetic "starbursts" tends to obscure the importance of the long-term, nearly normal stellar creation in gas-rich galaxies that may continue vigorously despite a collision. We derive a star formation rate for NGC 6240 and estimate that the efficiency of stellar creation is higher than that found for isolated galaxies, although this calculation has a large uncertainty. A normal formation efficiency indicates that new stars in the system would have been born anyway in the gas-rich precursors. We take the opportunity to comment briefly on another popular "ultraluminous" "starburst" galaxy, Arp 220, and apply some of the same techniques to analyze this system. As with NGC 6240, there is little evidence for stellar formation at a rate sufficient to explain alone the observed far-infrared luminosity, and the radiation from the older stellar population probably exceeds that available from young stars. However, as opposed to NGC 6240, Arp 220 appears to require major additional heating sources other than light from old and very young stars. As with NGC 6240, the derived efficiency of star formation in Arp 220 has a large uncertainty, but it is very roughly in the range of that found for isolated, nearly normal galaxies. We extend our detailed discussions of NGC 6240 and Arp 220 to other systems and attempt to quantify the role of enhanced interstellar absorption in merging galaxies; to derive the ratio of infrared to true stellar luminosity for a sample of normal, dwarf, and "superluminous" galaxies; and to point out our considerable disagreement with past work on this ratio. Large infrared-to-visual ratios in merging galaxies have been widely interpreted as indicating some unusual and very energetic process, but the ratios are as much a product of heavy interstellar obscuration as they are of luminous dust emission. In most previous work, the luminosity of the older stellar population in galaxies of all types and, therefore its potential for heating the dusty interstellar medium, has been systematically underestimated and often overlooked. Throughout our paper we emphasize the importance of proper treatment of the effects of a dusty interstellar medium. Uncertainties in the spatial distribution of galaxian dust are relatively unimportant for small values of the extinction, which is commonly assumed to be the case for normal isolated systems. However, for very dusty galaxies derivation of many source parameters requires knowledge of the relative star/dust distribution and a careful handling of the numerical approximations to the extinction. In our appendix, we briefly describe the effects of an alternative assumed dust distribution within galaxies. We take several opportunities to argue against the use of M82 as a prototype for objects such as NGC 6240, Arp 220, and putative "starburst" galaxies in general.
