01 / SEVERAL KINDS OF LOWERINGc.1400—1799
A downward word before a reliable curve.
The early record locates several senses; it does not support a continuous frequency story.
02 / PRINT VISIBILITY1800—1873
A new word enters print; melancholy still leads.
Depression becomes continuously visible in books, but it is not yet the dominant historical neighbour for low mood.
03 / A SHIFT IN VISIBILITY1874—1928
Depression crosses melancholy; market language gains weight.
Its printed visibility stays above melancholy from 1874 onward. That is a visibility crossover, not semantic replacement.
03A / PHRASE FREQUENCY
business depression
CORPUS WORDS1931 PEAK
04 / ECONOMIC CRISIS1929—1939
1932: the highest peak is not a diagnosis.
The core word rises sharply during the Great Depression period alongside independent economic phrase series.
05 / POST-CRISIS PLATFORM1940—1979
The peak recedes; diagnostic language stays marginal.
The broad word remains visible after the economic peak, while clinical labels have not yet formed a durable high-frequency band.
06 / DIAGNOSTIC BRANCHES1980—2022
Diagnostic labels rise; the shared word stays multiple.
Clinical phrases become more visible after the early 1980s. The core word still names more than diagnosis.
CLOSING FINDING
A branching word does not become one diagnosis.
This is not one straight path from sadness to sickness. Depression is a branching word: mood, market, pressure, landscape, weather and diagnosis share its spelling without sharing one meaning.
SOURCE / CORPUS
All plotted annual rates use the retained English Google Books Ngram series. Lexical anchors and contextual dates remain separate evidence layers.
METHOD / UNIT
Every charted book rate is the stored annual Google Books Ngram fraction multiplied by 1,000,000. The source file is the English corpus, 1500–2022, with source smoothing set to zero.
Period values are arithmetic means of every annual value in the labelled inclusive range. Raw annual plots remain unsmoothed unless the chart itself is labelled as a nine-year centered mean.
The 1874 crossover requires depression to exceed melancholy on a nine-year centered average for twenty consecutive years. Modern thresholds require five consecutive years above the stated rate using the same centered mean.
The 03A wheel compares business depression, financial depression and economic depression on one shared 0–3.5-per-million scale. Every phrase retains all 66 annual positions from 1874 through 1939.
Economic-event dates and controlled-vocabulary dates are contextual layers. They are not normalized with book-frequency values and do not prove causal relationships.
WHAT THIS CANNOT CLAIM
A single depression frequency series is not sense-classified. No chart treats emotional, economic, physical or clinical senses as shares of one whole.
Early Ngram values are sparse and potentially noisy. The first chapter therefore uses lexical anchors rather than a continuous early-frequency plot.
Phrase frequency is not a count of diagnoses, people, disease prevalence, economic output or the share of meanings carried by the core word.
RIGHTS / LINKS
Google Books Ngram Viewer data and all linked institutional sources remain subject to their own terms and attribution requirements.
Words Over Time presents derived rates, small rendered excerpts of source metadata, and links for research navigation; it does not republish source corpora.