This is a matter of knowing your data. If the spectra are from a spectral library taken in a lab, then they are often in percent reflectance that ranges from 0-1.0 (0-100%). If you see the data values are within the 0--1.0 range, then you can assume the library is not scaled and the scale factor would be 1.0.
Let's say your library was collected from a FLAASH or Quac corrected image. These tools also return data in reflectance but it is scaled by 10000, so your data may go up into the 1000s. Maybe you want to compare this library to lab spectra that range from 0-1.0. You would import your FLAASH/QUac scaled data using a scale factor of 0.0001. Or you could scale your lab spectra by 10000.
As for the units of the wavelengths - if the blue, green, red bands look something like [450, 550, 650] - these are nanometers. If they look like [0.45, .55, .65.], these are micrometers. Those are the most common units.
So the answer is... it depends on the data :)