Every Seven Years, or Not: Intervals Between Recorded Episodes of Pon Farr in 214 Vulcans in the Vulcan Science Academy Medical Archive, 2120–2190

1. Introduction

Pon farr is the periodic drive by which adult Vulcans are compelled to seek a mate. It is described as recurring about every seven years, and its onset brings a fever that, if unresolved, can be fatal (Marsden-Rell, 2370). The best-known outside account is of a Vulcan officer's episode aboard a Starfleet vessel in 2267. Vulcans themselves do not discuss the matter with outsiders, and the medical record of it is thin.

What we know of the interval comes from that thin record. The figure of seven years is used in Starfleet Medical's references and in comparative texts, and it is usually given as a rule, without a count behind it (Tennyson-Odum, 2372). The record it rests on has an obvious weakness. A physician sees only the episodes that come before a physician, and a private matter may be handled at home, in the family or by the healers of a temple. An episode that is not seen is not written down. A file with two recorded episodes 14 years apart would then not describe a 14-year cycle. It would describe a missed one.

A precedent exists. An earlier study in this journal found that most of the rise in recorded Pa'nar cases after 2154 reflected under-recording, since files before the reforms omitted a stigmatised disease that patients kept from physicians. That study used a neural series of the archive that was opened to outside scholars in 2373, and the archive holds further series. We ask whether a similar pattern affects the recorded intervals of pon farr. We write in 2376 from the archive's files.

2. Methods

The Vulcan Science Academy Medical Archive holds physicians' case files, and it was opened to outside scholars in 2373 (Vulcan Science Academy, 2373). We took every file from 2120 to 2190 in which a physician recorded at least two episodes of pon farr in the same individual, whatever the reason for the visit. We excluded 2154, the year of the reforms, as transitional, and we identified a clear entry of an episode by the Medical Council's later standard for recording cycles, applied retrospectively to the earlier files (Vulcan Science Academy, Medical Council, 2190). We assigned each pair to an era by the year of its second recorded episode, and we excluded nine pairs whose two episodes fell on either side of 2154. This gave 214 Vulcans, 72 with files before 2154 and 142 with files after it.

For each person we took the interval between the year of the first recorded episode and the year of the second. Years are those of the Federation calendar, and the archive records episodes to the year, so intervals are whole numbers. We used only the first two recorded episodes for each person, so that each person contributes one interval and no person's intervals are counted twice.

We described the intervals by their mean, standard deviation and median. We defined a long interval as one of 12 years or more, a cut chosen before the analysis at a point between the seven-year cycle and its double. We compared the proportion of long intervals between eras with Wilson intervals (Iskander-Wray, 2369), an odds ratio, and Fisher's exact test. We compared mean intervals with Welch's test (Rasmussen-Toure, 2368), for all intervals and for the shorter ones alone. All tests were two-sided at .05.

3. Results

Across the 214 people, the mean interval was 7.6 years (SD 2.2), the median was 7 and the range was 5 to 15. The interval was exactly 7 years for 76 (35.5%), and it fell between 6 and 8 years for 180 (84.1%).

The two eras differed (Table 1). Before 2154 the mean interval was 8.54 years (SD 2.88; 72 people), and afterwards it was 7.18 years (SD 1.49; 142 people). The difference was 1.36 years (95% CI 0.65 to 2.07, p < .001), and a rank-sum test agreed (p = .003).

Long intervals accounted for the difference. Twenty-two people had intervals of 12 years or more, 17 of 72 (23.6%; 95% CI 15.3 to 34.6) before 2154 and 5 of 142 (3.5%; 95% CI 1.5 to 8.0) after. The difference was 20.1 percentage points (Newcombe 95% CI 10.7 to 31.3), and the risk ratio was 6.7. The odds of a long interval were 8.5 times higher before the reforms (95% CI 3.0 to 24.1), a different figure from the risk ratio, and Fisher's exact test gave p < .001.

Those 22 long intervals averaged 13.41 years (SD 1.01), close to twice the mean of the 192 shorter intervals, which was 6.98 years (SD 0.89). Among the shorter intervals, the mean was 7.04 years before 2154 (55 people) and 6.96 afterwards (137 people), a difference of 0.08 years (95% CI −0.20 to 0.36; p = .58).

4. Discussion

In these files the underlying interval between episodes was close to seven years, and the recorded interval was longer when an episode had probably not been recorded. The shorter intervals were tightly grouped around seven and did not differ between eras. The longer ones clustered at 13 to 14 years, close to twice that, which is the pattern that would arise if a single episode had been missed between two that were written down. The longest interval was 15 years, so no interval is long enough to suggest two missed episodes.

The excess of long intervals before 2154 matches what the earlier study found for Pa'nar syndrome, where files before the reforms undercounted a disease that patients kept from physicians. The two cases differ in kind. Pon farr is not a stigmatised disorder, and as far as the files show, Vulcan reserve about it is older than the reforms. We do not know why recording improved after 2154. It may be that physicians and patients became more willing to discuss it, or that the archive's practices for entering cycles changed. The files do not say.

A practical point follows. A count of episodes taken from the archive will understate them, and the understatement is larger before 2154. An estimate of the length of the cycle drawn from the mean of all recorded intervals would be biased upward by about a year and a half for the earlier files (8.54 years against 6.98), and the shorter intervals give a better estimate of the cycle.

5. Limitations

We cannot count unrecorded episodes directly, and reading a long interval as a missed episode is an inference from its length. Some long intervals could reflect real variation in the cycle, and the archive says nothing of the Vulcans whose episodes never came to a physician at all. The sample is limited to people with at least two recorded episodes, so those with one or none are absent, and their cycle may differ. Only the first two episodes for each person were used, and later episodes may be recorded more or less completely. The 12-year cut point was chosen by us, and the results would shift slightly with another. The pre-reform and post-reform windows have different lengths (2120–2153 and 2155–2190), which truncates long intervals differently. One interval per person removes dependence within a person only. Bondmates or relatives seen by the same physician or temple healer may share recording habits, and the Welch comparison of a mixture of short and long intervals is descriptive, which is why we also report a rank-sum test. Intervals are in whole years, which limits the precision of the shorter intervals, and the small number of long intervals after 2154 (5) makes that comparison imprecise. The archive's files come from ShiKahr and its physicians. They may not represent Vulcans who were treated elsewhere. Finally, reasons for the change after 2154 are not recorded, and the reforms are a marker of the period and not a demonstrated cause.