Dr. Shreve: Thank you. In light of these comments there might occur to you questions that you would like to address to the panel members, and I think we could take five to ten minutes for this purpose. Floor: I disagree with Dr. Fleming's interpretation of diffusion method. I think we will be able to collect data effectively. Dr. Fleming: I hope you are right, John. I just say from past experience that I have had in this business I am just a little bit skeptical, particularly when you get to higher altitudes. question of Dr. Magar. I hope Jam wrong. I would like to aska I don't exactly understand your last statement that, no matter how small a sample you get, if you take your time you are going to be able to find out what it is. Dr. Magar: I didn't mean to take time, necessarily. I said that if you have a good diagnostic laboratory you can ana- lyze and keep on analyzing where you have the time to do this. Dr. Fleming: But isn't there a lower limit, given a specific technique for a specific application? Dr. Magar: My statement was not quite correct. If the sample is really too small, you might not have enough to per- form the analysis. Mr. Reed: I have a question on a subject that seems to have been overlooked. Are we fundamentally trying to get a few measurements of conditions at a few levels, and assume that this integrates uniformly over the world? There are arguments in favor of uniform distribution. On the other hand, there are strong possibilities of non- uniformity, and a few measurements may be considerably in error if you attempt to integrate them around the world. My ownfeeling is that it may be better to try to hold down the cost and sophistication of individual measurements so that we can make a tremendous number of observations at different places and at different times in order to make a better integration than is permitted with a few precise and sophisticated measurements, assuming worldwide uniformity. In that respect, it might be good to direct our rocket development toward a minimum cost so that we can operate on a synoptic scale. We have billions of weather observations stored in the National Weather Records Center, and from this we are utterly unable to give good weather forecasts. Iam afraid that a few measurements won't even begin to solve the problem for fallout prediction or future injection predictions. 162 We have to have a lot of data even to get any kind of a handle on the situation.