refinement before adjustment, 
refinement during ad justment , (method of self-calibration) 
refinement after ad justment, 
This experiment is part of systematic error refinement after 
adjustment. 
II. Basic Principles and Steps of a Posteriori Compensation 
For a posteriori compensation of systematic errors, first 
of all a fictitious model has to be selected, common points 
Of all models in the whole block or a part of the bloc::, 
according to coordinates ( Xm ,Ym,Zm) in System of model coor- 
dinates respectively, will be plotted to the corresponding 
positions in the fictitious model, such à fictitious model 
is called a reference model. A reference model according to 
a certain length can be devided into several small parts 
which are called sub-blocks. The average positions for all 
Common points fallen within a sub-block is called the core 
Of sub-block (or tie point). The method Of à posteriori com- 
pensation is the one by which a reference nodel iS established 
after a whole block adjustment, and by the use of the method 
of mathenatical Statistics, residuals of coordinates at com- 
MON points in models (so far as the method of independent model 
is concerned) are counted and analyzed, values of Systematic 
errors within individual core of Sub-block in reference model 
are computed. After T-testing,as siznificant systematic errors 
contained has been confirmed, then corrections for a posteriori 
compensation of systematic errors Should te added to nodel 
coordinates. After that a whole block adjustment will be per- 
formed once again and it Should be carried out again and 
again, until no Significant systematic error Will appear. 
The basie steps on a posteriori compensation are as follows: 
1. Acquisition of corrections for systematic errors at tie 
point 
After several adjusted iterations (two iterations are de- 
fined for this experizent), some residuals of model coordi- 
nates had been taken out of the positions of some tie roints 
(10 tie points have been used in this experiment) in accord-