IBM Core Memory Production
e-mail, Ken Shirriff to Len Shustek
You asked about core memory production and if it was automated. The book "Memories that shaped an industry" describes IBM's core memory production in great detail. Pages 154-159 discuss how the first core planes were wired by hand, but IBM rapidly developed ways to automate this process. IBM introduced a core threading machine in 1956 that cut the time to thread a core from 25 hours to 12 minutes. IBM continued to develop automation techniques.
I like this photo of a guy mixing up ferrite powder and looking like he's in a bakery:
By 1970, IBM was producing over 20 billion cores per year. (p243)
Pages 250-251 discuss how IBM was running out of factory space for manufacturing cores in 1964 to meet the S/360 demand. Plan manager George Tamke took some cores to Japan and found that they could make hand-wired core planes as good as the automated ones. IBM then employed thousands of people in Taiwan to wire core frames by hand, finding that it was a few dollars cheaper per plane than automated manufacturing, due to the low labor costs. IBM ended up doing most of the core stringing in Taiwan, while the US plants assembled the planes into arrays and memory units. Competitors without automation also moved manufacturing to Asia, causing IBM to lose much of its competitive advantage.
Ken