In spectroscopy, frequency is often expressed as wavenumbers, which are reciprocal centimeters (I have \DeclareSIUnit\wn{\per\cm} in my standard preamble). I would like to show the shift in frequency of a laser as a function of temperature, whose units should look like cm$^{-1}$/K but always either comes out /cm/K or cm$^{-1}$ K$^{-1}$. While I realize that all are technically mathematically correct (which is the best kind of correct), it is less ambiguous and customary in the field to distinguish between wavenumbers as a frequency and inverse length by writing the units as in the first case.

Is it possible to tell SIunitx, perhaps as part of the declare unit command, that the exponent should be treated as part of the unit always, even if the per-mode is changed?

  • 1
    Can you point to an online source where the requested output is used?
    – egreg
    Jun 2, 2014 at 17:01
  • The best example that I'm familiar with is the HITRAN spectroscopic database. The units for the intensity are given as wavenumbers per molecule per square centimeter. Mathematically, you could combine the the frequency and area to get a unit of per molecule per volume, but because the distances arise from two different physical quantities, you lose descriptive physical meaning by doing so.
    – craigim
    Jun 2, 2014 at 19:10

1 Answer 1


Define \wn slightly different using siunitx' \tothe or \raiseto:




A complete example:





enter image description here

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.