You may suspect a software identification error. The L lines of Bi and Th are in the same range. See:

http://physicsopenlab.org/2016/03/02/xrf-mineral-analysis/

If you are sure that Bi is absent, than you may consider that the "detected" Bi is in fact Th. Don't use the concentration values (you need a new calibration).

Cornel Radu is right.

Th Lalpha1 and Bi Lbeta2 coincide at ~12,96keV.

You may share your XRF spectrum. So we may see what is going on in your spectrum.

You may identify the emission lines by yourself:

http://www.med.harvard.edu/jpnm/physics/refs/xrayemis.html

When line superposition occurs, element identification in general is not done by one line, but via a set of emission lines.