169 
intestinal or other cells and could create transformation, transduction 
types of insertion of new genetic information into our cells. 
Thirdly, they could produce a disease by means of products, the 
enzymes, the cellulase, the example that was given yesterday, botulinus 
toxin, insulin, and an enzyme that changes a pathway in _Ej_ coli and makes 
a metabolite into a toxic metabolite. 
So this can impinge here, but these can cover almost all of the cases 
that one can think of. Now, there is a very crucial variable that I think 
ought to be brought into the guidelines soon, when we understand it better, 
and it is also crucial for evaluating risk. That is, as Paul Berg alluded 
to yesterday, a question of whether the _E. coli translates, and it is an 
insufficiently discussed variable. 
If coli translates the foreign DNA it can be a danger because of 
the regulation problem, particularly because it can make new pathways. It 
is irrelevant to the question of transferring nucleic acids, except it might 
facilitate the transfer, and of course the products are the big danger in a 
translating E^_ coli system. 
However, if it does not translate, the products automatically are no 
worry; the transfer of the nucleic acids is. And the question of regula- 
tion is minimized because it only changes regulation by inserting into a 
controlled area where the RNA is competing for ribosomal sites or some such, 
many fewer scenarios for producing bad results. 
Now, can one evaluate these hazards? I think the experiment of the 
polyoma in E_j_ coli or in lambda can evaluate this mechanism very effi- 
ciently. If the experiment is done with enough variables repeated enough 
times, it covers a lot of ground. In response to Dr. Goldstein, we will 
not tip this experiment in favor of negative results. We are not going to 
stop if we get a negative. We are going to keep on going until we are con- 
vinced that we won't get anything except a negative. We are leaning it 
toward getting positive results. 
So I think we can evaluate this risk scenario and very possibly elimi- 
nate it as a serious danger. I think this one also can be evaluated experi- 
mentally. I think if one takes shotgun clones from Drosophila or frog, or 
whatever are considered the best to work with, and characterizes 100 or 
1,000 of them for, Did they change the behavior or the pathogenicity of the 
K12 that they are in? One can at some point say okay, this is a rather 
ridiculous mechanism; we don't have to worry about it. 
So I think this can be made a set that we don't have to worry about. 
I don't think we can evaluate these, because each is a case of its own. 
But this is a tremendous step, because a nontranslating system is useful 
for two of the three major uses of this technology. You can clone with it 
[310] 
