I'm not a native English speaker so the use of the word "rough" was probably not adapted. I'm more or less within a 2% range with my FTIR method compared to the results from external certified laboratory for protein analysis (most generally Kjeldahl or Dumas method).
I work on a sample type that is a complex heterogeneous matrix, but that does not vary that much in terms of molecules concentration (lipids, proteins, hydrocarbons, etc... are fairly distributed - we could say it's like having 100 cow milk samples).
It is possible that my latest samples are lower in concentration than previous ones, but i got about 10% lower estimation with my standards as well. They are starting to get old but i keep them in a dried dark place.
I checked the wavelength, and there is no major shift. I got a 3 cm-1 difference for my right and left edges which is, I believe, not a full explanation of the 10% difference.
But what's odd is that I did those analysis again, after the last weekend, and this time I found the same results (+- 0,3%) with my 2 analyzed standards, and a value for my unknown sample 9,8% above the one I considered to be wrong.
I also contacted our supplier which say sunlight and/or humidity is in no case causing trouble to analysis since the instrument has been designed to work properly even in subtropical climate.
I proceeded to do weighing instrument verification, and it seems that the whole deviation was coming from the balance.
I have a technician coming in the next days/week to do some check-up and repairs.
I guess that in the meantime I will have to weigh my samples with an another precision balance.