yard or basement lab," said Mike the
Angel.

Fitzhugh nodded emphatically. "Exactly. We can't let that technique out
until we've found a way to keep people from doing just that. The UN
Government has inspection techniques that prevent anyone from building
the conventional types of thermonuclear bombs, but not the pinch bomb."

Mike the Angel thought over what Dr. Fitzhugh had said. Then he said:
"That's not all of it. Antarctica is isolated enough to keep that
knowledge secret for a long time--at least until safeguards could be
set up. Why take Snookums off Earth?"

"Snookums himself is dangerous," Fitzhugh said. "He has a built-in
'urge' to experiment--to get data. We can keep him from making
experiments that we know will be dangerous by giving him the data, so
that the urge doesn't operate. But if he's on the track of something
totally new....

"Well, you can see what we're up against." He thoughtfully blew a cloud
of smoke. "We think he may be on the track of the total annihilation of
matter."

A dead silence hung in the air. The ultimate, the super-atomic bomb.
Theoretically, the idea had been approached only in the assumption of
contact between ordinary matter and anti-matter, with the two canceling
each other completely to give nothing but energy. Such a bomb would be
nearly fifty thousand times as powerful as the lithium-hydride pinch
bomb. That much energy, released in a few millimicroseconds, would make
the standard H-bomb look like a candle flame on a foggy night.

The LiH pinch bomb could be controlled. By using just a little of the
stuff, it would be possible to limit the destruction to a neighborhood,
or even a single block. A total-annihilation bomb would be much harder
to control. The total annihilation of a single atom of hydrogen would
yield over a thousandth of an erg, and matter just doesn't come in much
smaller packages than that.

"You see," said Fitzhugh, "we _had_ to get him off Earth."

"Either that or stop him from experimenting," Mike said. "And I assume
that wouldn't be good for Snookums."

"To frustrate Snookums would be to destroy all the work we have put into
him. His circuits would tend to exceed optimum randomity, and that would
mean, in human terms, that he would be insane--and therefore worthless.
As a machine, Snookums is worth eighteen billion dollars. The
information we have given him, plus the deductions and computations he
has made from that information, is worth...." H