Abstract: In order to obtain reliable adjustment results of densification networks,it is neces-sary to detect the blunders in the coordinates of junction points. In this paper,the coordinates of junction points are considered as correlated observations with covariance matrices,and the data snooping method is used to detect the blunders in the coordinates. Based on the theory of reliability and discernability of extended Gauss-Markov models,the localizability of the blun-ders in the corrlated coordinates is formulated,and demonstrated with the examples of typical densification networks (including GPS baseline networks). Some suggestions for the design of densification networks have been made. This study has also found that the value of a redundan-cy number, usually used as a measure for the reliability,may be beyond the interval0, 1 for correlated observations.