Ullica Segerstrale: I have a question I'd like to address to everybody on the panel. It was suggested by Bob Filler and again by Janine Larsen: What should we do about students fudging experiments in laboratories. It's easy to understand that it's not going to pay off, so to say, in the long run, because one has to learn to stand on one's own feet. But in the short run, it might pay off. Lab reports are graded on whether they are correct or not. So, how do we get around that?

Bob Filler: In the report that I cited from R & D this week, one of the people complained, "That's the root of all of our problems, the university. When we did our laboratory work, there were too many experiments. We had just two months. So we worked together. One group of us did a, another group did b, and so on. When we were finished, we traded reports. We just learned whatever they had done, nothing more than that:" The truth of the matter is, it requires a tremendous amount of time, effort, and concentration by professors and honest assistants to teach the virtues of integrity to students. They will have to learn from example.

I'm not here to attack any faculty member. They are honest hardworking people. But, many of the things you pick up in life are picked up much earlier. Whether it is in childhood, in school, in the university. No matter how well you're taught, however, there are some individuals, for reasons of their own, who feel they have to cut corners, fudge, and the like. Let me tell a story.

Twenty-five years ago at Purdue University, a well-known professor whom I know had many of students. One student was brilliant. The professor wanted this student to work on a new chemical compound. Nobody had seen anything like it. Soon the student reported, "I produced the compound. It possesses all the properties expected:' They published the results as a Communication in the Journal of the American Chemical Society, the top ranked journal in chemistry. The student graduated and went off to work for Syntex in Mexico City.

Then someone from Italy read the paper and said, "That's baloney. Theoretically, this is all wrong." The professor at Purdue said, "I can't believe that" He assigned a new student to repeat the experiment. The student couldn't reproduce it. The professor assigned a second student to reproduce the experiment. He also failed.

I told you the professor's prize student was brilliant. He was. He had dry-tabbed the whole thing. He figured out in his head what the properties of the material should be, fudged everything through, got his degree, and went off to Mexico: "Yes, I did this, so what? I got my Ph.D., so long:" The professor had to publish a retraction in the journal, red-faced as it was, and at the next Purdue commencement, the student was defrocked, in absentia.

Hassan Nagib: What Bob Filler has told us happens. It is simply a disgrace. It makes me feel almost unclean if it happens in my profession. I think what is clear is that the failure comes from the reward system. For example: You have a material and you want to measure its conductivity at various levels-maybe some kind of a semi-conductor or something like that. Every student in the class is expected to do that experiment for different temperatures and have a graph with 15 points and many pages of notes on them. That is what drives the students-and drives the professors in their own funded programs-some things that are on the edge.

What we have to be creative about is that reward system: For example, if every student in the class only had to do the experiment at one temperature, but each had to get the accuracy to two decimal points instead of one, and had to share the results with the class, each would have to trust every other. Each of them would have to take the 14 other points and put them on the same graph. They wouldn't have to be there until 6:00 p.m., although the class is supposed to end at 4:30 p.m. They would not have to miss dinner in the dorms or something like that. We have to be creative with our reward system. What do you get for what you do-not just finally in the product but in the learning process? I think that is the key. One quality data point is better than fifteen so-so data points.