the parlour clock, correct on tuesdays

the clock on the parlour mantelpiece is correct only on tuesdays. the rest of the week, it is wrong; not consistently wrong, not by any fixed amount, but wrong in the way a clock is wrong when its hands have been moved at random and left to drift. on a tuesday, beginning at midnight and lasting until midnight, it agrees with the kitchen wall-clock, with the keeper's pocket-watch, with the church clock in town, and with the sun. on any other day of the week, it does not.

what is known: the clock is a small french carriage-clock, gilt and bevelled glass, dated 1869. it has been on the mantel since the parlour was the parlour. the mechanism is sound; two clockmakers have examined it and found nothing to repair. the pattern is reliable from at least 1898 onwards, the date at which the volume's editor began keeping a log; whether it held before that, no one can now say. the clock has been correct, on a tuesday, regardless of whether the tuesday was a leap-day, a midsummer, or a sabbath fallen on a tuesday.

three accounts. the first holds that the keeper, who is fond of tuesdays, sets the clock secretly each tuesday morning and lets it drift the rest of the week; the keeper denies this and has been observed not setting it. the second holds that the clock's spring has a weekly resonance which, by some accident of manufacture, peaks every seventh day; this is the position of the second clockmaker, who is satisfied. the third, the editor's, is that the clock and the days of the week have a private arrangement we are not party to, and that the clock should be left to honour its arrangement. the eight thirty-six may or may not be relevant.

see · the clock room tea for tuesdays a clock clocks eight thirty-six