How Much of Your Roster Runs in the Dark

A 06:00 shift in Oslo starts before sunrise on 227 days a year. The same shift in Singapore starts before sunrise on all 365. If that ordering is the wrong way round from what you expected, it is worth understanding why, because the reasoning behind it decides which of your sites actually has a darkness problem.

Most writing about daylight and work is about seasonal swing: the long northern winter, the short northern day, the well-known misery of arriving and leaving in the dark. That is a real phenomenon and it is not the only one. Absolute darkness at a given clock hour is a separate problem, and the tropics have it in a permanent form that no season ever relieves.

Two different problems that both look like darkness

Near the equator the day is close to twelve hours long all year. Singapore's sunrise sits within a few minutes of 07:00 every day, which means an early shift beginning at 06:00 begins in darkness in January and in June and in every month between. There is no summer during which the problem goes away, and equally no winter during which it gets worse. It is a constant.

Move north and the constant becomes a curve. Oslo's sunrise runs from around 03:55 in June to around 09:15 in December, so the same 06:00 start is comfortably in daylight for a third of the year and deep in darkness for the rest. The annual total of dark starts is lower than Singapore's, but the experience is worse in the specific weeks where the curve bottoms out, and the swing itself is the thing people notice.

Which of those two your operation has determines what you can do about it. A constant is a design decision: if the 06:00 start is always dark, the question is whether it needs to be 06:00 at all. A curve is a seasonal plan: the roster is fine for half the year and needs attention for the other half.

Longitude moves the whole curve

Latitude decides how much the answer moves through the year. Where you sit inside your time zone decides where the whole curve sits, and the effect is large enough to reverse intuitions. Madrid is at 40 degrees north, well south of Oslo, and a 06:00 start there is before sunrise on all 365 days, because Spain keeps Central European Time while sitting far to the west of the meridian that time zone is built around. Its clock runs ahead of its sun by roughly an hour and a half.

Anchorage, at 61 degrees north, records fewer dark 06:00 starts than Madrid does. Latitude alone would never predict that. This is the same mechanism behind the finding that an extra hour of natural evening light reduced sleep by 19 minutes at US time-zone borders, where two nearby places keep the same clock but see the sun at measurably different hours, and where income on the later-sunset side was about 3% lower outside cross-border commuting zones.

The practical version is that a country name is a poor proxy for light exposure and a time zone is a worse one. The relevant comparison is between local sunrise and the actual start time on the roster, which is what counting the dark shifts in a roster is for.

The clock change moves people across the line

Sunrise moves gradually. The clock does not. Twice a year in most of Europe and North America, the shift stays where it is on paper and jumps an hour relative to the sun, which can take a start time from the daylight side of sunrise to the dark side overnight. Plotted across a year, the daylight band has two visible steps in it, and they fall in late March and late October, when the sunrise line is already moving faster than at any other point in the year.

That compounding is why the transition weeks are the ones worth watching in absence and lateness data. The change is not subtle in the way the underlying astronomy is: it arrives in a single night, and it arrives on top of a period when the sun was moving quickly anyway.

What darkness is known to do, and what it is not

Here the honest answer is narrower than the popular one. Among 32 indoor day workers in Kiruna, at 67.86 degrees north, winter sleep onset came 39 minutes later than in summer, mid-sleep shifted later, weekly sleep was 12 minutes shorter, and sleepiness was higher. Greater morning light exposure was associated with earlier mid-sleep, and greater evening exposure with later timing. Those workers had ordinary electric lighting, so the seasonal difference persisted despite it.

That is a study of sleep timing and sleepiness in 32 people. It did not measure work output, and no study cited here does. So the defensible operational claim is that seasonal daylight changes sleep timing and sleepiness among indoor workers, and that any step from there to efficiency has to be tested rather than assumed.

This matters because a great deal of published material converts darkness into a productivity percentage, and those numbers do not come from studies of darkness. We keep the full read on latitude, daylight and worker energy with the sample sizes attached, which is usually the detail that goes missing.

The seasonal cognition claim is a misreading

One claim recurs often enough to be worth killing on its own. You will read that sustained attention peaks near the summer solstice and working memory near the autumn equinox, usually presented as a reason to schedule analytical work in autumn and collaborative work in summer.

It comes from a single study of 28 healthy young adults, each kept for four and a half days in an environment stripped of seasonal cues before being tested in a scanner. The seasonal peaks it found are peaks in brain activity. Performance on the same tasks did not change significantly across the year. The result is about how much neural resource was recruited to hold performance steady, which is genuinely interesting and is not a reason to move anyone's work to a different quarter.

Seasonal differences in cognition itself have been found, in a much larger analysis of 3,353 older adults, where cognition was higher in summer and autumn by an amount equivalent to 4.8 years of age difference. That is a different population answering a different question, and we set the two side by side in the evidence on seasonal cognition.

What to actually do with the count

A dark start is an exposure, not an outcome. Its value is that it is the part of the problem a roster can change: nobody can move sunrise, but a start time is a decision, and knowing that it lands before sunrise on 227 days makes the size of that decision visible in a way that a sunrise table never does.

• Count it per site, not per country. Two locations in the same time zone can differ by an hour of sunrise, and two locations at the same latitude can differ by more than that.
• Separate the constant from the curve. A site that is always dark at 06:00 has a start-time question. A site that is dark for four months has a seasonal plan question.
• Watch the transition weeks. Clock changes move people across the sunrise line in one night, on top of the fastest-moving part of the year.
• Measure locally before designing around it. Lateness, absence and incident rates are already in your systems, and a winter baseline is worth more than any published seasonal curve.
• Protect morning light where the roster allows. It is the intervention the mechanism most directly points at, and it is cheap. It is not a promised return.

None of this makes darkness a solved problem. It makes it a measured one, which is the difference between a roster that acknowledges where it is in the world and one that treats 06:00 as the same instruction everywhere.