The CMB is measured through a fog of molecular lines, and the fog's thickness is uncertain at the 1-to-10 percent level in exactly the channels precision cosmology uses. Carbon monoxide's rotational ladder (115, 230, 345 GHz and beyond) radiates from every molecular cloud in the Galaxy and from unresolved extragalactic populations, landing directly in the primary frequency bands of Planck and ground-based CMB experiments. Planck's component separation produced the first all-sky CO maps as a byproduct, and the lesson was uncomfortable: CO leakage contaminates temperature maps at levels that vary with the detector bandpass, the cloud velocity structure, and the line ratios assumed, with residual contamination after cleaning estimated between 1 and 10 percent of the CO signal depending on region and channel.
The deeper problem is "dark" molecular gas: CO-faint H2, gas whose carbon is in atomic or ionized form or whose CO is photodissociated, which Planck dust analyses and gamma-ray studies indicate may rival the CO-bright component in mass, organized in extended envelopes and high-latitude cirrus precisely where CMB analyses prefer their masks thin. Underestimating molecular structure at high latitude propagates into temperature and, more dangerously, polarization analyses, where dust and CO correlate, contributing to the foreground complexity that already complicated one famous B-mode claim. As CMB-S4-era sensitivity targets primordial signals at the noise floor of foreground modeling, line contamination misjudged at the percent level stops being cosmetic.
The standing is a workmanlike systematic with sharpening stakes: the CO sky is better mapped each year, the dark-gas inventory keeps growing, and the question is whether the foreground complexity converges before the next generation of primordial claims depends on it.