To the Editor: We have recently encountered a minor discrepancy between the literature documenting the POINTER program and the program as it is implemented. This program allows the segregation analysis of nuclear families, conditioned upon pointers, who are relatives outside the nuclear family through whom it was ascertained. Pointer codes consist of a single digit followed by a letter, where the digit indicates the degree of relationship, and the letter defines more exactly the relationship of the pointer to the family to whom it points. Lalouel and Morton [1981] state that a pointer is "assigned only to the closest component of the nuclear family." However, as the program is implemented, when the pointer is related to the parents through a child the relationship should in fact be specified between the pointer and the parents, and not between the pointer and the child. In contrast, in the program COMBIN [MacLean et al., 1983, 19841, which also implements the pointer logic and coding, the relationship for a pointer through a child should be given for the child, and not the parents.
Thus, following the pointer coding scheme [Morton et al., 1983; MacLean et al., 19831, if a family is ascertained because one of its children is a parent of the proband, the proband should be included as a pointer with the code 1D for COMBIN and 2D for POINTER. COMDS, a program recently developed from COMBIN, which considers severity and diathesis information [Morton et al., 19911, follows the COMBIN coding scheme.
A worked example is given in Figure 1. The conditional likelihood for a fully recessive gene with complete penetrance is easily shown to be P = q(l + 3q)/4, or 0.154 for prevalence of q2 = 0.1 with corresponding displacement t = 3.239. The RA option in POINTER and COMDS gives -2 In P , from which the values in Table I may be calculated. The result for 1D in POINTER corresponds to q(l + q)/2, as if