Bob: I was executing the following program from cards (1-80) using single-step, starting at 29: ....5...10...15...20...25...30...35...40...45...50...55...60...65...70...75...80 5022I9CM004079B036 B036 H089000H0890S0#007079Y2U01|0YYYYYYYYYYYYYYYYYYYB029 1 1 1 1 1 1 1 1 11111111111111111111 1 ,002009,017024,028035)008015)022029)036036NNNNNNNNNNNNNNNNNNN1002 1 1 1 1 1 1 1 1 11111111111111111111 1 After three times through the very clever loop from 36-79, the instruction at 76-79 was modified to 5015 (RP 15). When I executed that in single-step mode, the next card was read, then a "Unit not attached error" was generated (because I had not attached CDP). I decided to ignore the error and proceed using "s". By then, 1-80 had been changed to the second card's contents, so the instruction that had been "RP 15" had been changed to 1002 (R 2). i1401 proceeded to execute this instruction, instead of branching to 15. I don't quite know what changes to make to cause i1401 to finish executing RP 15 instead of starting over and executing R 2. This was part of a "core-to-card" program written by Ken Dritz in 1964, which IBM published in their General Program Library as 1.4.158. The cards leading up this point are ,008015,022029,036043,050057,076080,058059,060061,075066,,,,,,,,NNNNNNNNNNN5050 D100120DDDDDDDDDDDDDDDDDDD1050 Y100120DDDDDDDDDDDDDDDDDDD1029 1036I9CM004079B036 H089000H0890S0#007079Y1Y01S0YYYYYYYYYYYYYYYYYYYB022 5036I9CM004079B036 H089000H0890S0#007079D1Y01S0DDDDDDDDDDDDDDDDDDDB022 1036I9CM004079B036 B036 H089000H0890S0#007079D2U01|0DDDDDDDDDDDDDDDDDDDB029 5022I9CM004079B036 B036 H089000H0890S0#007079Y2U01|0YYYYYYYYYYYYYYYYYYYB029 ,002009,017024,028035)008015)022029)036036NNNNNNNNNNNNNNNNNNN1002 Van