Flags at Anchor, California Ship Registry

Between 1769 and 1821 the ships anchoring in California changed nationality. Spain arrived alone, and within fifty years most vessels in Spanish harbours flew another flag. This is that change, year by year, from the registry's own visit records.

Figure 1. Share of recorded arrivals by flag, every year from 1769 to 1821, as a five-year centred mean. Each year's own unsmoothed share is available on hover and in the table. The line breaks wherever its window falls below twenty records, and the strip beneath gives n for every year, so no stretch of the curve is read without the evidence under it.

Why share, and why smoothed. Raw arrivals per year rise and fall with how densely a year happens to be documented at least as much as with how many ships came: 1797 alone holds 65 recorded visits, largely one exhaustively catalogued stretch of San Blas supply runs, while four years in this window hold none at all. Counts would draw that survey pattern as history. Share is robust to an uneven denominator, and a five-year centred mean is robust to a thin year, but both are still computed on the same uneven evidence, so the annual points and the annual n are printed underneath rather than smoothed away. Where the window holds fewer than twenty records the line simply stops.

Who arrived first, and on what authority. Six first visits, marked in the lane beneath the plot and listed with the registry's own attestation in the table: Spain 1769, the founding voyages at San Diego (inferred, draft); France 14 September 1786, La Pérouse's Boussole and Astrolabe at Monterey (stated, reviewed); Britain 14 November 1792, Vancouver's Discovery into San Francisco Bay (stated, reviewed, exact date); the United States 29 October 1796, the Otter of Boston under Ebenezer Dorr at Monterey (reviewed); Russia 8 April 1806, the Juno under Rezanov and Khvostov at San Francisco (one of only thirteen rows in the registry at verified status); Argentina 20 November 1818, Bouchard.

Two of those need their qualifications stated. The American first is 29 October, not the offshore sighting of 28 May 1796 that stands earlier in the file, a sighting fifteen leagues out is not an arrival, and the registry reclassified it as such in August 2026. And nothing British precedes Vancouver: Bancroft's own chapter is headed “Vancouver's First Visit,” and he writes that “the first intercourse of the Californians with subjects of a foreign power was with the French under Jean François Galaup de La Pérouse in the autumn of 1786,” calling it “the first visit of a foreigner to California.” The near-miss is real but is not a British visit, the Princesa Real that touched Monterey in September 1790 was the British sloop Princess Royal, seized from Britain at Nootka in 1789 and sailing under Manuel Quimper and Spanish colours. A British hull, two years early, under someone else's flag.

Where each line starts. A centred five-year mean spreads every value backwards across half its window, which in an earlier draft of this figure lifted each flag off the axis two years before it arrived: Russia read as 1804 when the Juno came in 1806, the United States as 1794 when the first American arrival is 1796, Britain as 1790 when Vancouver's Chatham came in 1792, France as 1784 when La Pérouse came in 1786. First arrival is the whole point of a series like this, so it is not smoothed: each line is held flat at zero until the year of its first recorded arrival, and that year is ticked on the axis in the flag's own colour.

Why the chart ends in 1821. Because the data cannot honestly carry the Mexican period yet. Forty-three visits dated 1822 or later are flagged Spanish in the registry, among them vessels named Morelos, Matamoros and Mariquita, which are Mexican, and Brookline, Volunteer and Leonidas, which are American. Thirty-nine vessel names in the file appear under more than one flag (Eagle is Spanish, Russian and American; Cossack is Spanish and American), and 19 per cent of post-1821 rows carry such a name against 7 per cent before 1822. Some of that is real, ships were sold and re-flagged, and the Ayacucho's registry is genuinely contested, but most of it is coding noise, and the registry's own method note already lists mis-flags among its standing false-positive classes. A composition chart of Mexican California would be measuring that noise. It waits on a flag audit.

What the figure says

The crossing happens under Spanish rule. Spain holds every recorded arrival until the 1780s and four in five through the 1790s. Then the line falls through the 1800s and the American line passes it, while California is still a Spanish province, a decade before the Mexican war of independence and forty years before the conquest. The empire lost its monopoly on its own harbours long before it lost the coast.

Russia appears from nothing and takes a third of the traffic. Not a single Russian arrival before 1806; within a decade Russian vessels are more than a third of recorded arrivals, peaking above forty per cent in the 1810s. The Ross venture is legible as a step change in a shipping series.

Britain is a spike, not a presence. The British share peaks in the mid-1790s, Vancouver's surveys and the Nootka aftermath, and then subsides. It is an episode, where the American share is a trend.

France is small, and shown anyway. Seven records across three separate occasions, La Pérouse in 1786, a French matter in 1808, Roquefeuil's Bordelais in 1817, which is why the French line is three short segments (1786–88, 1806–10, 1815–19) rather than a curve. Two of those seven are flagged for review. Argentina is Bouchard alone and appears only as an arrival. Both are kept visible rather than folded into a grey "other": a category with two records is still a category, and burying it hides which two.

A line is drawn only where its flag was actually there. It begins at that flag's first recorded arrival, and it stops again wherever the flag is absent for two years or more. A flat line along the axis draws absence as though it were a measurement, and the eye reads it as “here, at nil” rather than “not here”, which is why France appears as three separate episodes rather than one long line at zero, and why nothing runs beneath Britain before 1792 or Russia before 1806. A single isolated zero inside a continuous presence is kept, because there it really is a measurement. The arrivals themselves sit in their own lane below the axis, out of the plot they would otherwise crowd.

A caution on the last years. The Spanish line does not fall to zero because the supply system never fully stopped, but the window's final years are also its thinnest: the war of independence cut the San Blas run and the record with it. Read the right-hand end against the n strip, not on its own.

Table view, every value in both figures

On the form. The design is a plain time-series of composition, the oldest form in the field: William Playfair drew the first of them for England's trade balance in the Commercial and Political Atlas of 1786, and the elements here, labelled axes, a recessive grid, one line per series, direct labels at the ends, are his. Two later rules govern the rest. Edward Tufte's: spend the ink on data and take it off everything else, which is why the grid is a hairline and no line carries a number at every point. Tufte's other one is that a chart should show the data it was computed from, met here by the n strip beneath the plot and by the table, which carry every annual value the curve averages, rather than by scattering fifty-three years of points across the plot itself. Colours are not chosen by eye. The four series hues were run through a colour-vision check in both light and dark mode: the closest adjacent pair (Britain against Russia) separates by ΔE 9.1 under simulated protanopia and ΔE 22.9 under normal vision, against thresholds of 8 and 15. Two of the light-mode hues sit below a 3:1 contrast ratio on this surface, which is permitted only where the reading never depends on the colour, hence the direct label on every line and the full table above. Dark mode is a separate set of steps chosen for the dark surface, not an automatic inversion of the light one.

Source: California Ship Registry, data/visits.csv · derived series data/flags-by-decade.json (built by scripts/build_flags.py, which is deterministic, rerun it to reproduce every number here).