Genotyping using melting temperature data

I am trying to assign genotypes to individuals from a population using melting temperature information. I am fairly certain that all individuals have one of two genotypes and will therefore have one of two true melting temps. I can cluster the measured temps for all individuals into two groups, but do not know how to determine how certain I am that a single measured melting temp is a measure or estimation of a single actual or true melting temp.

In other words, how can I determine the confidence that a specific measured value is an estimation or measurement of a specific true value?