Observe that we can bisect a day into two ⚀clock periods (semi·days), each 6 ⚀dwells long (equivalent to 60_z ⚀breathers, or 600_z ⚀trices, or 10_z|12_d hours, or 500_z|720_d minutes). But where should we make the partition? The modern Western convention, of course, is to divide the day at noon and midnight, yielding the familiar Ante Meridiem (A.M.) and Post Meridiem (P.M.) ⚀clock periods. These are Latin for “Before Mid-Day” and “After Mid-Day”.

However, some cultures (notably the Abrahamic religions) have traditionally bisected the day at sunrise and sunset, yielding Daylight and Nighttime ⚀clock periods. In Latin, these might be called Lux and Nox. The traditional approach of these cultures is to observe the actual appearance and disappearance of the sun across the horizon, marking sunrise and sunset on a daily basis. This makes the division between Lux and Nox, and the lengths of both, a complex, varying function of latitude, longitude, season of the year, Daylight Savings Time, and the effects of twilight atmospheric refraction.

For that matter, the actual points of “mid-day”, when the Sun is at zenith, and “mid-night”, when the Sun is at nadir, are just as subject to daily and seasonal fluctuation. For simplicity however, Primel will assume “nominal” or “average” times for these events, with “mean sunrise” and “mean sunset" deemed to occur exactly half-way between “nominal midnight” and “nominal noon”.

But which way should we divide the day? Interesting question. Dozenal base provides easy divisibility by 4—so what if we bisected the day both ways? Then this would divide the day into quarters, analogous to the four seasons of the year or the four phases of the Moon. Primel proposes calling these quarters ⚀phases of the day. Each would be 3 ⚀dwells long (equivalent to 30_z ⚀breathers, or 300_z ⚀trices, or 6 hours, or 260_z=360_d minutes).

We can give each ⚀phase of the day a distinct name. One way to do this is simply to use the cross-product of the names of the two orthogonal bisections. But it turns out that English already has suitable names for these periods: 

Hence the ⚀clock periods are combinations of two adjacent ⚀phases each:

We can divide each ⚀phase into 3 ⚀dwells, and assign them names: Early, Mid, and Late:

In a Primel world, people would likely tell time in terms of the ⚀trice count. Since the ⚀trice is 5/6 of a minute, this gives accuracy to the minute and better, using only three dozenal digits. Conventional 12_d-hour clock time requires four digits plus an indicator of the current ⚀clock period, A.M. or P.M. TGM time in ⚀lapses does better in that the hour is encoded in a single digit and the top digit encodes the  ⚀clock period (0=A.M., 1=P.M.). However, one more digit only gives accuracy down to the nearest 5-minute (6-⚀trice) ⚀block. To get at least minute accuracy in TGM requires 4 digits. But because the ⚀lapse is half the size of the ⚀trice, and only 25_d seconds long, this provides too much accuracy, at the expense of taking up an additional digit. Bottom line, the Primel scheme is the most compact.

The so-called “9-to-5” job would start at 460_z ⚀trices, half-way through Mid Morning ⚀dwell, and would end at 860_z ⚀trices, half-way through Late Afternoon ⚀dwell. This gives a total duration of 400_z ⚀trices or 4 ⚀dwells (8 hours). Then again, in a Primel world, businesses might opt to start and end the work day on round ⚀dwells, let's say 400_z to 800_z ⚀trices (8:00_d A.M. to 4:00_d P.M., or Mid Morning through Mid Afternoon), or perhaps even 500_z to 900_z ⚀trices (10:00_d A.M. to 6:00_d P.M., or Late Morning through Late Afternoon). In the latter case, this might be referred to as a “5-to-9” job, standing the conventional term on its head.