Thanks Micha.
Can you tell me how to specify stochastic model for directions which are we assume to measure in 3 gyruses (in my Excel I divided 3" with √3 (square root) and for distances we assume to measure twice every direction (in my Excel I divided 2 + 2ppm with √2, ppm part depends on approximate distance value). That is probably reason why I get different results. Can you also tell me why is a posteriori uncertainty zero value for directions?
If you have least square root concepts that you implemented in software, please give me link.