Number-word frequency in the King James Bible

How often each number is used in the King James Version of the Bible, counted from the spelled-out numerals rather than from digits. The KJV writes its numbers as words, including compound forms such as "six hundred threescore and six", so a count that searches for "666" finds nothing and a count that searches for "six" finds the wrong things. The corpus (Project Gutenberg ebook #10, header and licence stripped) is parsed with a compound-numeral evaluator: adders accumulate, "hundred" scales, "thousand" banks, and -score survivals fold in (threescore = 60, fourscore = 80). Whitespace is collapsed before matching, because the source wraps lines mid-phrase. Distributive constructions such as "two and two" are counted as separate uses rather than summed. Cardinals are the primary series; ordinals are counted separately as a robustness check. Number-words thin out as the numbers get larger, as they do in most prose. Against that slope, seven and twelve stand out and forty does not. Seven is used 376 times, about 5.5 times the mean of its immediate neighbours (six, 103; eight, 33). Twelve is used 156 times, about 8.9 times its neighbours (eleven, 21; thirteen, 14). Forty is used 93 times, about 1.1 times the decades either side of it (thirty, 80; fifty, 83). Ten is included as a control rather than as a result. At 176 uses it is about 9.3 times its neighbours, the largest ratio in the set, but base-ten counting inflates ten and its multiples in any language spoken by people with ten fingers, so its prominence is the expected behaviour that shows the method working. Seven and twelve are not round numbers in base ten and rise anyway. The standalone value 1 is parsed but excluded from the series. In English "one" is also a pronoun and an article ("no one", "one of them"), so its raw count of 1,941 measures English grammar rather than a number's use. It is still parsed so that compound numerals resolve. This measures a 1611 English translation, not the Hebrew and Greek beneath it. Biblical Hebrew writes numbers as words rather than numerals, and a translation's choices affect every count here. A frequency is also not a meaning: this records how often a translation reaches for a word, and says nothing about what the number meant to whoever wrote it. Files: bible-number-frequency.json (a metadata block plus one object per number: number, cardinal, ordinal) and bible-number-frequency.csv (a commented header carrying the corpus caveat, the exclusions, the DOI and the licence, then number,cardinal,ordinal). 24 rows each. Both are byte-for-byte identical to the copies self-hosted at https://numeluna.com/bible-number-frequency.json and .csv. The underlying corpus is public domain. What is licensed here under CC-BY 4.0 is the parse and the tally, not the text.