Each of the cells has a top contact along one edge and a bottom contact all over its base, so we were able to assemble the 12-cell groups like shingles, with the bottom edge of one cell covering the top edge of the next, leaving only the active area of each cell exposed. But this meant that each module would be over four inches long and only 14 thousandths of an inch thick—far too weak to withstand stress and vibration. To support the cells, we decided to mount them on a metallized ceramic base. But this presented a problem: If the cells were soldered directly to the base, the different thermal expansion rates of the silicon and the ceramic would cause the structure to break during the cycles of extreme changes in temperature that Telstar would pass through. We remedied this by connecting each cell to the ceramic support by a thin U-shaped strip of silver (_see above_). Since silver has a much higher thermal expansion coefficient than silicon, we added tiny sandwiches of Nilvar or Invar (36% nickel, 64% iron) where the cells were attached. With this mounting method, the cell modules withstood thermal and mechanical shocks much more severe than those they would undergo in actual use. In one test, for instance, an entire cell module with its cover plates was first dipped in hot water, then plunged into liquid nitrogen at a temperature of -195° Centigrade. In orbit, the temperature range for the satellite was not expected to be more than from +80° to -100°C, with a rate of change of no more than three degrees a minute. Finally, we needed to find the right kind of transparent protective cover for the Telstar solar cells, both to keep micrometeorites from damaging the sensitive and very thin diffused layer and to slow down the incoming electrons to nondestructive energy levels. For micrometeorite protection, only a thin layer of hard transparent substance was needed; for electron protection, the cover plates should have a mass of no less than 0.3 gram per square centimeter (as we explained above). And there were two other important considerations: The material we used should not be darkened or discolored by prolonged exposure to ultraviolet radiation, and it should have good thermal conductance, so that some of the heat absorbed by the solar cells could be conducted out to the cover plates and re-radiated. All these requirements led us to the choice of clear, man-made sapphire. Although sapphire is more expensive and difficult to make than the equivalent quartz or glass, it only has to be 30 mils (three hundreds of an inch) thick. Twice this thickness would be required if quartz or glass were used.