case. But in the case of the better quality protection-the best one I have discussed-it would certainly be necessary for the population to decontaminate if they came out of the shelter after a heavy attack, say, 2 weeks following that attack.

Mr. HOLIFIELD. Those figures are figures representing exposure? Mr. HANUNIAN. Yes. This indicates the amount of exposure, the fraction of the total dose that does get through.

I would like now to go on to the final attack comparison that I want to show.

а

Earlier, in the course of exploring how fatalities among the general population are affected by changes in the weight of military or urban center attacks, we noted that an attacker intent on destroying large numbers of people may be able to do it very cheaply. We have seen that diverting only a few hundred megatons from military targets to cities will produce perhaps 50 million deaths from prompt effects alone, if the population is unprepared, and that with fallout the total might be as high as 80 million. But we have also seen that far fewer deaths would result from an attack directed solely against military installations.

Thus it is clear that the validity of any forecasts of attack outcome depends on how well those forecasts reflect enemy motivations. Since these are somewhat unpredictable, prudence demands that we develop the implications of alternative motivations.

I have already made a gesture in that direction by considering attacks against cities and against military installations as alternatives between which the attacker may choose. There would be some point to considering variants of both sorts of attack, but variations of military attacks have the more interesting implication for war plans and for strategy, and it is these we shall consider here. Since some of these attacks may be at least as appropriate to a late time period as to an early one, we ought to allow for the possibility that changes will be made in our civil defense program.

Therefore, I shall now expand the variety of the assumed civil defense postures to include, as one alternative, the use of the specially constructed shelters.

Mr. HOLIFIELD. Now, on this last column there, the fallout shelters, are you including in that category shelters in basements which have been adapted to give the maximum protection from fallout and other types of shelter which might be arranged underground; are you? Mr. HANUNIAN. Yes.

What I mean to imply by this assumption about the residual radiation reaching the population is that this may be the result of there being improved basements, carefully improved basements, with extremely careful use being made of those improved basements.

Mr. HOLIFIELD. Do your figures appear just for houses that have basements, or are you including in addition to basements-is this a figure for the whole population providing they have the equivalent of underground protection which could come from an adapted basement?

Mr. HANUNIAN. Let me try to clarify what I have in mind here. The assumption I have made is really that a certain amount, a certain fraction, of the radiation gets through to the population.

This assumption is consistent with alternative interpretations. One such interpretation is that the people are spending their time continuously in well-improved dwelling basements, that is, basement shelters with sandbagged walls.

Mr. HOLIFIELD. During most of the attack, for the period of time necessary for the radiation to decay on the outside?

Mr. HANUNIAN. The blast survivors enter these basements of theirs, where they have specially prepared shelters in the corner of the basement, perhaps with sandbagged walls and with a roof of beams, and sandbags on the top of that.

They spend every minute of the day there for the first couple of days, and virtually no time outside of those areas for the next couple of weeks.

Now, the assumption I have made is also consistent with a shelter which requires less of the population by way of careful behavior, that is to say, it is consistent with a much better quality shelter (a more expensive shelter, to be sure, but a better quality shelter) that gives them a higher degree of protection during the time they are in the shelter, and, therefore, allows them to spend a larger fraction of their time outside the shelter without getting any larger dose than they would have by that very careful behavior in the basement.

Mr. HOLIFIELD. It is evident there are many basements-let us take the tunnels under the Capitol, for instance, and other rooms under the office buildings in the Capitol, that would be much superior to an ordinary small basement in a residential area from the standpoint of protection

Mr. HANUNIAN. That is right.

Mr. HOLIFIELD (continuing). Because they are deeper, they have more over head in the basement and all that sort of thing.

Mr. HANUNIAN. Yes. There are two important variables, really, that determine the amount of effective shielding. One is the quality of the structure sheltering the people; the other is the quality of their behavior-how well they make use of that shielding.

REASONS FOR DIFFERENT ENEMY ATTACKS

It is well known that surface bursts produce many times the amount of local fallout that air bursts produce, weapon for weapon. An attacker will not be strongly motivated to use only surface bursts unless his intent is to maximize the fallout hazard. If he uses surface bursts rather than air bursts, he sacrifices military effectiveness so far as blast damage is concerned. There is, to be sure, some compensatory advantage gained in that the resulting fallout will make more awkward the continued use of any target installations that have escaped physical destruction.

It is only against fairly hard structures, structures vulnerable to nothing less than about 100 pounds per square inch peak overpressures, that surface bursts appear to be advantageous for him; and thus we would not be justified in assuming that an attacker would use surface bursts universally. He might very well use them against hard targets;

he might choose to attack hard targets in that way. But for the rest we cannot say. And we are, therefore, obliged to consider air bursts as well.

There is another basis for differentiation among military attacks: The allocation of attack weight between missile launch sites and other prime military targets.

While contemplating target sets, we noted that these two categories have distinctly different locational patterns, missile sites being in the west, the other military installations being distributed more or less uniformly. We can expect that casualties among the population at large will show some dependence on where the attack emphasis falls. Further, it is not at all clear on which category the Soviets would concentrate the bulk of their attack. We can expect this to change with time and with circumstances. (For example, imagine a Soviet planner pointing out that if their forces failed to achieve surprise in attacking our missiles, our missiles might have flown from the silos, leaving behind them profitless targets. The planner might urge that retargeting be considered for this contingency.)

CASUALTIES FROM VARIOUS 3,000-MEGATON ATTACKS ON MILITARY TARGETS

I want now to consider four combinations of these variations: missile launch sites receiving alternatively one-sixth and five-sixths of the entire attack weight, with surface bursts being used either universally or only where physically hard structures are the targets. Let us now examine the outcome of such attack variants.

Here, once again, fatalities are shown as a percentage of the national population. Now, however, we consider only total deaths, not blast and fallout deaths separately. (See fig. NH-5.)

FIGURE NH-5.-TOTAL DEATHS RESULTING FROM ATTACKS ON CONUS MILITARY

The various attacks are arrayed along the base here. Notice that all the attacks are alike in that they involve delivery of 3,000 megatons total yield.

The two clusters of bars at the right represent the outcome to be expected if the attacker uses surface bursts exclusively, that is, this cluster of bars, and this one. The two clusters of bars-I am sorry; I mean to say that one of the two clusters in each pair relates to an attack allocated principally to missile sites. The other principally to other military installations. Within each cluster the four bars capped in different colors show how the population fares under the four different civil defense assumptions that I have described. The

shaded cap, the colored cap, on each bar reflects the extent of uncertainty in outcome deriving from those same two weapons characteristics I have referred to before: the fission fraction, and the radioactivity level.

It is when the attacker uses surface bursts exclusively (the two clusters of bars at the right of fig. NH-5) but especially when he directs his attack mainly to military installations other than missile sites-it is in these cases that we find maximum fatalities, and this is true whichever civil defense assumption we use. Notice how a shift of the attack to/from missile sites other than military installations reduces the level of the fatalities.

Turning our attention now to the other clusters, the other pair of clusters, we see that fatalities are lower still if airbursts are used against some of the targets. Shelters are, however, nonetheless very helpful in further reducing population loss; there is a substantial falloff, as you see.

Now, with some weapons being airburst, it is when the brunt of the attack is delivered to ICBM sites that fatalities are heavier.

When surface bursts are used universally, it is the other way around. The explanation is not hard to uncover. Other things being equal, we should expect that an attack directed mainly at our missile sites would produce relatively few deaths. These sites are mostly in sparsely populated areas, so there are fewer people nearby to be affected, especially by fallout. And this is what the chart, the righthand half of figure NH-5, shows.

It is the opposite half, the left-hand half, that needs explaining. Here the shift in the direction of attack is accompanied by another shift; other things are not equal. I have assumed (plausibly, I think) that the Soviets would use surface bursts against sites that are hard, and missile sites are virtually the only hard targets. Thus this pair of clusters at the left reflects more extensive use of surface bursts, along with the shift to heavier emphasis on attacking ICBM sites.

The result is that much more radioactive material has been deposited in the United States, and the fatality levels are consequently higher.

ESTIMATES FOR 10,000 AND 30,000 MEGATON ATTACKS ON THE UNITED

STATES

The final charts, identically arranged, reveal that a similar variability and similar progression of results obtain in the face of much larger attacks in attacks delivering 10,000 megatons and higher. Fatalities, of course, are naturally higher with the higher yield. (See figs. NH-6 and NH-7.)