From:  IBM's Early Computers by Charles J. Bashe, Lyle R. Johnson, John H. Palmer,  and Emerson W. Pugh; The MIT Press, 1986,  page 374-375.

       Although Wood had no background in solid-state devices, he concluded he could not devise successful transistor circuits unless he made his own transistors.  Lacking any facilities for processing materials  he purchased germanium crystal diodes, broke off the covers, carefully leaving the two leads of each attached, and with the help of a microscope attempted to attach the necessary third lead.  The resulting transistors were similar in structure and function to those first made at the Bell Telephone Laboratories.

      Known as point-contact transistors, such devices consist of two thin wires in point contact with the surface of a piece of germanium and a third wire attached to the base.  One of the point-contact leads is called the emitter because it is used to emit minority charge carriers into the base region of the transistor, that is, electrons into p-type or holes into n-type material.  The other point contact is called the collector 'because it collects minority-charge carriers from the base.'  Point-contact devices failed to achieve significant use in computers because they depend on a difficult-to-control metallurgical formation where the collector is attached to the germanium. Junction devices, the theory of which was published by Shockley in 1949, ultimately provided the stable characteristics needed for large, transistorized computers. 9   But point-contact devices were discovered first and were the basis of all early efforts to develop and use transistor circuits.

      Wood says he "fumbled around a lot and learned a lot" making his own point-contact transistors but did not make much real progress until he hired a young assistant with the Ph.D. in physics from MIT. 10  {Geoffrey Knight Jr.}   Working together, they found that probes made of silver and gold wires did not give current gains greater than one but that a certain copper-alloy wire did.  Merely touching the copper-alloy wire to the germanium crystal was not sufficient, but if a large current was passed through the collector wire in contact with the crystal, the wire bonded itself to the germanium and created a transistor with good current amplification.  This bonding process was called burn in or  forming, a metallurgical process, poorly understood at the time, that gave point-contact transistors their unique characteristics.  Wood and his assistant correctly speculated that the copper-alloy wire contained elements that entered the germanium and helped create the desired structure."

      Most of what they learned about the mysteries of transistor forming was presumably already known to those who had invented the transistor at Bell Laboratories; but such information was not yet generally available, and in such a new field there was always the chance of unexpected discoveries leading to new inventions.

      By the spring of 1950, Dickinson's group had made substantial progress in the fabrication and use of transistors.  The fabrication process began with 1N48 germanium crystal diodes purchased from the General Electric Company.  These were chosen because they had "a satisfactorily high impedance in the reverse voltage or back direction" and because they exhibited "a livelier surface." The group soon developed techniques for optimizing the placement of both the emitter and collector leads instead of using the lead attached by the diode manufacturer as one of the transistor point-contact leads.  A transistor triode test unit was constructed that automatically plotted the collector current versus its voltage for six different values of emitter voltage (and emitter current for later versions) on the screen of a cathode-ray oscillograph.  " Using this test unit, a technician in Wood's group - who also served as elevator operator and chauffeur for T. J. Watson, Sr. - could electrically form transistors in steps, testing the device after each step and adjusting the position of the whisker (metal wire) contact to achieve the desired transistor properties." Later it was found that the transient currents of the curve tracing system were so large that they were sufficient to form the contacts automatically as the technician moved the electrodes over the surface of the germanium.