Notes. ^(a) Total column density of H₃⁺ estimated by summing the population of the measured H₃⁺ levels in this table. We did not correct this total column density from the ∼10–50% H₃⁺ molecules in unobserved metastable levels, which depends on the physical conditions and excitation of the H₃⁺ clouds (Appendix D and Oka et al. 2019). The upper limits are derived from the upper limits of the three lowest metastable levels of H₃⁺: (1,1), (1,0), and (3,3). ^(b) Adopted column density for the H₃⁺ levels (1,1), (1,0), (2,2), (2,1), (3,3), and (3,0). We averaged the columns from the different branches, excluding, in most cases, those derived from the R branch due to the dilution by the stellar continuum of the R branch (Sect 2.3). We also excluded some absorptions partially blended with emission lines in some of the objects (e.g., Q(3,3) in Arp 220). ^(c) H column density estimated from the extinction affecting the 3.4–4.4 μm continuum (see Sect. 3.1; Donnan et al. 2024). The estimated uncertainty is ∼0.3 dex. ^(d) H₃⁺ fractional abundance. The uncertainty is dominated by the 0.3 dex uncertainty on N_H. ^( * ) In NGC 3256-S the P(2,2) transition is detected (Table B.3 and Fig. C.1), but its EW is affected by the broad CO₂ ice and gas absorptions. For this reason, we adopt the upper limit estimated from the Q(2,2) transition. ^( † ) For IRAS 14348 SW, the P-branch column densities are 2.3±0.6 times larger than those from the Q branch. Thus, we considered the Q-branch values for the column of the individual levels, which are used for population ratios, and obtained the total column density by multiplying their sum by 2.3.