The absorption spectrum produced by a 12 cm cell of C${\mathrm{O}}_{2}$ gas, and the Bunsen flame emission spectrum, to ${3}_{\ensuremath{\mu}}$ were recorded by means of a new self-registering infrared quartz prism spectrograph of an auto-collimation type. A new weak absorption band was found at ${2.02}_{\ensuremath{\mu}}$ which forms a linear, though not a harmonic, frequency series with known bands at ${2.73}_{\ensuremath{\mu}}$ and ${4.25}_{\ensuremath{\mu}}$. This was identified with a ${1.99}_{\ensuremath{\mu}}$ flame emission band, which had been assigned by Paschen to water vapor. The ${2.7}_{\ensuremath{\mu}}$ emission region was widely resolved into two components of practically equal intensities, with maxima at ${2.58}_{\ensuremath{\mu}}$ and ${2.76}_{\ensuremath{\mu}}$, corresponding to maxima identified by Paschen with water vapor and C${\mathrm{O}}_{2}$, respectively. When the quartz prisms were replaced by a similar rock salt pair the ${4.4}_{\ensuremath{\mu}}$ emission occurred as a nearly symmetrical band, although Barker's measurements with high dispersion showed it to be double. Calculations of the Bunsen flame temperature from his observed separations gave 2000\ifmmode^\circ\else\textdegree\fi{} K, in rough agreement with previous pyrometer determinations.