Quarterly port activity — Algeria (2016-2022)

Traffic through Algerian ports, quarter by quarter, 2016 to 2022. 439 observations across nine headings: total traffic, hydrocarbons, non-hydrocarbon traffic, container tonnage and count, passenger traffic, vessels entering, average wait at anchorage and average berth stay.

Each heading is split into landed and loaded, with its total. Algerian ports handle some thirty million tonnes a quarter, of which about eighteen are hydrocarbons.

Passenger traffic shows its Mediterranean seasonality without needing comment: 90,000 passengers in the first quarter of 2018, 450,000 in the third. And the 2020 collapse — 1,125 passengers in the third quarter, against 466,949 a year earlier — measures the border closure exactly.

Three identities, on different axes. Landed plus loaded equals the heading's total; hydrocarbons plus non-hydrocarbon traffic equals total traffic; and the published growth rate is indeed the change between the two amounts. The third rests on the column the first two ignore: a reading shifted by one row still satisfies the first and fails here. Eighteen tables out of twenty-two pass, and the delivered file satisfies 91 identities out of 91 — after arbitration between bulletins, which was not a given.

Each quarter is published twice. That quarter's bulletin gives it as the current figure, then the following year's gives it again as the comparison. The two readings are set against each other: two bulletins agreeing on the same tonnage did not err the same way. Where they differ — 27,715 then 27,744 for the first quarter of 2021 — the later reading prevails, and a column flags it.

What the layout demanded, and what ignoring it cost. This domain stayed out of our catalogue a long time, for want of being readable. The cause came down to one character: the Arabic unit annotation leaves its parentheses behind, and "27 339" arrives as "27 339 )". The cell is no longer a number, the whole row collapses into its own label, and every heading lost its total — precisely what makes the table verifiable.

Two rows in the series — wait at anchorage, berth stay — also have their cells offset by one column in typesetting: "1.1 0.99 -9.32" would read as a negative waiting time. Nothing settles whether it is an offset or a typo, and guessing would be worse than keeping quiet: those rows are set aside without the table being refused.

The growth rate is not published, though it served as a check. It recomputes exactly from the two amounts, and shipping it would add a column not to be mistaken for a value.

Units and source. Thousands of tonnes, counts, days and thousands of dinars, by heading and as published. Four tables are refused because their sums do not add up — a container total of 107,108 where its own components make 207,108. We do not correct published values; we refuse the table and say so.

This product reads alongside our "Transport and tourism", which follows the same bulletins for road, rail and air.