This sample of Home Owners' Loan Corporation (HOLC) loans contains 3,032 observations from New York, New Jersey, and Connecticut. This data was originally used in an NBER study, "History and Policies of the Home Owners' Loan Corporation," by C. Lowell Harriss (1951). This sample was taken by the NBER in 1947 with a sample from three warehouses in New York City. These three warehouses were the central repositories for all records for New York, New Jersey, and Connecticut. In two warehouses, the NBER collected information from every thirtieth loan, and in the third, every sixtieth. The Harris (1951) book states "Owing to the inaccessibility of records, only every sixtieth loan was sampled at one of the three warehouses where the HOLC records were stored. As a result, it was necessary to inflate the loan sample from that warehouse, raising the total number of loans actually sampled from 3,172 to the inflated number of 3,883." There seem to be two possible interpretations of this passage. First, an additional 711 loans were sampled from the warehouse in question, at some other date or by some other unknown method. This seems unlikely, because the "inaccessibility of records" would seem to indicate that a further sample was not forthcoming, and also because there is no evidence from the microfilming process that more than 3172 records were ever present. This leaves a second option, that observations from the warehouse in question were "inflated" by increasing their weight in the original 1951 study. I latter interpretation is more plausible. The data were collected from three warehouses in Manhattan, including two on Park Avenue, and one on Christopher Street. The one-thirtieth samples were taken at the two Park avenue warehouses, yielding 1,158 and 1,303 observations respectively. The one-sixtieth sample was taken at the Christopher Street warehouse, yielding 711 observations. These added together total 3,172. The original sample was taken in 1947 and then microfilmed in 1948. The microfilm contained 3,172 slides, which was a 2.5% sample of loans from these three states. The difference is due to faulty microfilming, in which 140 observations were made unreadable due to overexposure. During the microfilming, 140 observations were lost, as noted above. The number of surviving loans, then, are 1,111 of 1158 from the first Park Avenue warehouse, 1,279 of 1303 from the second Park Avenue warehouse, 642 of 711 from the Christopher Street warehouse. These sum to 3,032.