Operation Castle, Project 4.1
E. P. Cronkite, V. P. Bond, R. A. Conard, C. L. Dunham and others; Naval Medical Research Institute, Bethesda, and U.S. Naval Radiological Defense Laboratory, San Francisco · WT-923 · October 1954 · declassified and released through the Department of Energy
The document at the centre of the accusation that the United States used the people of Rongelap as experimental subjects. It is cited constantly and read almost never — usually at second hand, from an encyclopaedia summary or an advocacy database. It runs to 94 pages, and its own first page of text states three objectives, of which the third is the one the argument turns on.
Open the PDF — 94 pages, 4.9 MB
Hosted here in full. Nothing has been re-compressed, re-OCR’d or altered.
What is on which page
A caveat that belongs at the top. This is a 1954 typescript scanned decades later. Its text layer is poor: it reproduces some pages cleanly and reduces others — particularly tables and figure captions — to noise. Passages are quoted below only where the scan is legible enough to read a sentence with confidence. Where it is not, the content is described rather than quoted, and said to be described. Page numbers refer to PDF pages, not the report’s printed pagination, because the printed numbers are among the things the scan loses.
PDF p.11 — Chapter 1, “Objectives”
The report states its own purpose, and the third objective is the whole controversy. “Project 4.1 was organized with the following specific objectives: (1) To evaluate the severity of the radiation injury in the human beings exposed to the fall-out radiation. (2) To provide for all necessary medical care for these individuals. (3) To conduct a scientific study of radiation injury in human beings.” Medical care and scientific study are listed as co-equal aims of the same project, in the project’s own words. Neither the defenders nor the accusers of Project 4.1 usually quote all three together.
PDF p.11 — “General Description of the Exposed Groups”
Who was under the fallout, in the report’s framing. “Shot 1 of Operation CASTLE was detonated on 1 March 1954. Following the detonation significant amounts of radioactive materials fell on the following populated neighboring atolls: (1) Rongelap, (2) Ailinginae, (3) Rongerik, and (4) Utirik.” (The scan renders the opening as SIN)I 1 of O]wration CASTLE; the reading is unambiguous but the characters have been normalised, which we note rather than pass over.) The report notes that the Rongelap group received the highest calculated dose and that they were “living under relatively primitive conditions in lightly constructed palm houses” — a sentence doing dosimetric work (thin walls stop little) and carrying a colonial judgement at the same time.
PDF p.12 — Table 1.1
The table the “exposed and control groups” phrase comes from. Table 1.1 is headed Exposed and Control (Unexposed) Groups and totals 267 exposed and 222 controls, the latter including a Marshallese control group of 117 from Majuro (the table sets “Marshallese, Control” against “Majuro Group” in adjacent cells). This is the passage usually paraphrased as the Marshallese being “divided into exposed and control groups,” and the distinction is worth stating precisely: the control group was unexposed and served as a measurement baseline. Nobody was exposed in order to fill it. What the table does show is that within days a fallout accident had been given the architecture of a study.
PDF p.12 — decontamination
A detail worth reconstructing, and labelled as reconstructed. Clothes were removed and laundered and skin and hair washed repeatedly with fresh water and soap. The report then records a difficulty specific to the islanders: that the radioactive material was hard to wash out of their hair because of the heavy coconut-oil dressing they used. This passage is among the worst-scanned on the page — the text layer renders “Marshallese” as M;{rshxll(isc — so the sentence above is a reconstruction from legible fragments rather than a quotation, and is deliberately not presented in quotation marks.
PDF p.21 — Section 2.3, clinical management
The reasoning for withholding antibiotics, which cuts against the simplest reading of the file. The report sets out four reasons for not giving prophylactic antibiotics, among them that “the premature administration of such drugs would not only possibly obscure medical indication of treatment, but might potentially lead to the development of drug resistant organisms in an individual with an already lowered resistance to infection,” and that the situation was not strictly comparable to the agranulocytosis produced by potentially lethal doses. Whatever else this document is, that passage is clinical judgement about treatment, not observation in place of it.
PDF p.35 — Section 3.2.4, Histopathology
Biopsies. “Seven biopsies were taken of neck lesions and one of an axillary lesion during the third to fourth week after exposure on the Rongelap people… A second group of biopsies from the Rongelap group (repeated in three individuals) were taken; four at the seventh week post-exposure and five at the eighth-week period.” Tissue was cut from burned people, more than once from some of the same people, in the weeks after the exposure. The report does not give a therapeutic rationale for taking them. What it gives is a selection criterion — “Most of these biopsies were taken from individuals who showed lesions of greater than average severity” — and a safety precaution: “Biopsies were not taken from open lesions or from the feet for fear of infection.” It records that “All biopsy wounds healed rapidly with no secondary complications,” and it records no discussion of consent.
What the document does not contain. We searched the full text layer for consent, informed and volunteer. There are no occurrences. That is not by itself evidence of concealment — a 1954 military medical report would not be expected to discuss consent, and the modern framework did not yet exist — but the absence is the fact on which the later ethical judgement rests, and here it is checkable rather than taken on trust.
Provenance and licence
Released by: the U.S. Department of Energy, through its OpenNet declassified-document system. The report was produced for the Atomic Energy Commission by the Naval Medical Research Institute and the U.S. Naval Radiological Defense Laboratory.
Original copy: OSTI OpenNet, document 0402776, CIC Marshall Islands collection. We did not obtain this document; we host a copy so the citation in the case file resolves to something durable. OpenNet addresses are opaque numeric identifiers carrying no title, which is a large part of why this report is cited at second hand.
Licence: Public domain — work of the U.S. federal government.
Classification: originally classified; officially declassified and released by DOE. This is the released version, not a leak.
How this copy was checked: Downloaded from OSTI OpenNet; 5,158,800 bytes, SHA-256 274c8d61977228f8c7bb01916d81ce8c05ad48c1fd674bb485ebd4169c3ee64b, byte-identical to the hosted copy. Opened and confirmed 94 pages, all 94 carrying some text layer, though quality varies sharply. Page 1 reads “OPERATION CASTLE — FINAL REPORT PROJECT 4.1 / Study of Response of Human Beings Accidentally Exposed to Significant Fallout Radiation,” by E. P. Cronkite, Commander, MC, USN, with ten named co-authors, October 1954. Embedded metadata gives the title, author “CRONKITE E P”, date 10/1954 and collection “CIC MARSHALL ISLANDS”. Every passage presented in quotation marks above was located, and read, at the PDF page given. Two passages are explicitly marked as reconstructed or normalised rather than quoted, because the scan will not support a verbatim reading of them.
94 pages · 4.9 MB · PDF