This measurement is given as a fraction of the wavelength of the light used to conduct the test. To have any meaning for a telescope intended for normal viewing, the test wavelength, l, should be 550nm,- where human vision peaks.  Some telescope manufacturers try to cheat by quoting the wavefront error obtained from using a more favorable wavelength for this test, to get a better looking specification.

A wavefront error of l / 4 = 550e-9 / 4 * 1000 = 137.5 micrometers. At the mirror's surface this will become 68.75 micrometers. Likewise, a wavefront error of l / 24 will be 550e-9 / 24 * 1000 = 22.92 micrometers, which will become 11.46 micrometers at the mirror's surface.  The smaller figures compensate for the fact the measurement counts both ray entrances and exits to and from the mirror, effectively doubling the actual imperfections in the off mirror measurement. Multiplying by 1000 conveniently converts the results from nanometers to micrometers, or microns if you prefer. I figured my curve to an accuracy 6 times better than required for Diffraction Limited Optics. I'm glad I signed my work.