APPENDIX H 
Nevertheless, the Working Group considers it essential that there 
be some experimental estimation of the "worst possible case. " We 
can very conservatively estimate that the worst possible case (specified 
-1 
below) will actually arise in less than one of ten (10 ) x -sensitive 
strains encountered in nature. This brings the required value for 
-4 
test (2) to 10 We therefore recommend that the following "worst- 
case" tests be included and show: 
(a) The number of fragment -containing bacteria per adsorbed 
-4 
input fragment -containing phage should be less than 10 and 
(b) The number of fragment -containing rearmed (wild-type) phage 
-4 
per adsorbed input fragment -containing phage should be less than 10 
-3 
Values of less than 10 for each of two genetic safety features one 
located to the left of the cloned fragment and the other to the right, 
are also acceptable. 
It is understood that the host in this worst possible case should 
be a non-permissive lysogen bearing a heteroimmune prophage which 
has base -sequence homology with the input phage and a compatible 
late gene regulatory system. 
D. The product of reversion frequencies of "disarming" mutations 

should be less than 10 
Comment : 
This test should be done in a non-permissive, non-lysogenic bacte- 
ria. Separate determination of the reversion frequencies of individual 
Appendix H — 4 
