I spent the last few hours working through Rolex’s latest patent filing, and as I dug into the drawings and claims, several questions emerged. What follows is my attempt to make sense of what Rolex filed, and to discuss whether or not this might be the groundwork for their first modern perpetual calendar.
Calendar complications frustrate watchmakers in many ways. There’s the aesthetic problem of “dragging systems” where the date crawls from 31 to 1 over several minutes or hours, and these look bad at midnight. Then there’s the mechanical complexity problem, where instant-jump systems require loads of intermediate wheels, levers, and programming cams which all fight for space in the case. There’s also the energy problem, because calendar mechanisms sapping power from the movement can cause amplitude drops. And of course, there’s the thickness problem, because extra components stacking up tend to add to the movement height.
Then there’s also user experience issues like forbidden zones where you can’t adjust the calendar, correctors which need careful handling, and the perpetual question (lol) of whether the added complexity is worth the maintenance headaches down the line.
That said, Rolex is proposing a system that jumps instantly but also keeps the mechanism (relatively) simple. Whether this layout is actually “innovative” or just a competent execution of existing ideas… I am not well-versed enough on calendar movements to say for sure. Anyway, let’s look into it, starting with the first diagram.
This is the complete assembly as you’d see it from the dial side. At the centre we have the date wheel (1) with its 31 teeth visible around the periphery. The date hand (13) is meant to point to the date, currently pointed at 12:00.
This is the underside of the previous diagram, and here you can see the energy accumulator (4) on the left, with a spring-loaded lever (41-42) and roller (43) riding against cam (33). The day wheel (5) sits at the top left, and the intermediate wheel (6) meshes with the date wheel’s teeth. The month wheel (2) sits at axis A2 on the right.
This sets out the overall spatial arrangement, where the date wheel sits centrally on axis A1 and everything else orbits around it.
This was the more baffling image for me... We know the date mechanism sits at the centre on axis A1, and it is made up of a wheel (11) with 31 teeth (1a) going around (I literally counted them to be sure!). Mounted on top of this wheel is lever (12), which pivots about its own axis A12 on the date wheel itself. A couple of things happen here; tooth (12a) on one end gets pushed by finger 32, pin (12b) on the other end contacts the month cam, and the lever body with slot (12c) that limits its movement range.
The date wheel also has a fixed pin (11a) that works with a little slot to restrict the lever’s pivoting amplitude to about 7° (approximately 0.5mm of movement at pin 12b). This retractable lever is how the system distinguishes months with different lengths.
Next is the month wheel on the right which sits on axis A2. This consists of a pinion (21) with 12 teeth (2a), month programming cam (22) with circular portions (22a) and radial projections (22b), and a month display disc (23) which is not visible in this image but shown in Figure 1. The cam’s projections are correspond with short months (22b) and the circular portions or smooth sections (22a) correspond with long months.
Finally, there’s the drive wheel (3) which sits on axis A3 and makes one complete revolution every 24 hours. Notice disc (37) carrying fingers 31 and 32; finger 31 shifts the date wheel’s 31 teeth (1a) directly. Finger 32 interacts with lever 12’s tooth (12a).
Beneath disc 37 we have cam (33) for energy storage, and the roller (43) from the energy accumulator (4) rides on this cam.
This is a useful exploded view of the drive wheel assembly, which clarifies the architecture considerably. From top to bottom, we have disc (37) at the top, carrying fingers 31, 32, and 38, the unidirectional coupling (35-36) connecting disc 37 to wheel 34, cam (33) for accumulating energy, and wheel (34) at the bottom with external teeth meshing with the movement’s hour wheel.
Wheel 34 is the gear interface with the base movement, and that’s what seems to rotate once every 24 hours. The number of teeth on wheel 34 depends purely on the gear ratio needed to achieve 24-hour rotation when meshing with the hour wheel.
My understanding here is that the number of teeth on this wheel are decided based on gear ratio requirements, not perpetual calendar programming. The wheel must rotate once per day regardless of tooth count.
I saw some suggestion online that this might lead to a perpetual calendar, but for this to be a perpetual calendar from the start, there must be a wheel that at least tracks the four-year leap cycle. This wheel typically has 48 positions (or equivalent) and makes one complete revolution in 48 months. I simply cannot find an additional wheel with 48 teeth that rotates once per 48 months.
The patent text says, roughly: “The cam may be an annual month cam (12-month cam, making one complete revolution in 12 months)” which implies the month cam (22) makes one revolution per year. For a perpetual calendar function, you’d need an additional 48-month wheel, which isn’t explicitly shown or discussed in the patent wording.
Further detail below1 because this technical stuff is getting out of hand for the main body 🤣
Annual or Perpetual?
The drawings only really show an annual calendar mechanism. The month cam (22) has programming for distinguishing 30-day months from 31-day months since you can see the four projections, corresponding to September, April, June, and November.
February would need manual correction at the end of the month because this system would try to advance from 28 to 29 (or 29 to 30), and you’d need to use a corrector to jump forward to 1 March. There’s no visible 48-month wheel or four-year cam in any of the diagrams, and the month wheel makes one revolution per year, according to the patent explanation.
And yet, in typical Rolex fashion, they take the piss! The patent also repeatedly says the system can be “simple or annual or semi-perpetual or perpetual”.
In essence, Rolex filed this patent repeatedly making claims without the required design, just so that they can later argue for intellectual property protection for all these variants. They will argue they are not speculating, but without the actual diagrams and drawings, there is no proof of their achievement, and they’re literally just “claiming” they’ve worked out how to extend this architecture to perpetual calendar function. 😂
I am no watchmaker, but in practice, the architecture in the current drawings could be extended to perpetual calendar (according to a couple of watchmakers who helped me with this) by:
Adding a 48-month wheel geared to the month wheel at 4:1 ratio. This wheel would make one revolution in four years, and track the leap cycle.
Mounting the month cam on the 48-month wheel instead of directly on the month display disc… Or having the 48-month wheel carry an additional cam that modifies the month cam’s ultimate profile.
More complex cam profiling on cam 22. Instead of just “projection versus circular,” you’d need multiple levels or profiles to create all the different jumps2.
Possibly additional fingers on drive wheel 3, or variable positioning of lever 12 to create different advancement amounts depending on what the programming cams allow.
This feels like Rolex has patented a calendar architecture that’s fundamentally modular and scalable. The drawings show the annual calendar configuration, which is simpler and easier to illustrate, but they have written the patent to try and protect this design as “a family of calendar mechanisms” which I think is quite cheeky (but I am no expert on patent law either!).
If this is allowed, then I suppose it is rather clever patent strategy. You file one patent covering a basic invention and also protect all other variants? If that’s actually how this works, it’s quite unfair because the claims in here are broad enough to prevent competitors from using many similar approaches at any complexity level.
Anyway, putting aside the annual versus perpetual debate, and patent strategy, what makes this filing interesting at all?
For starters, Rolex currently offers the Sky-Dweller which is an annual calendar, but the Sky-Dweller uses a different setup. The month is indicated by twelve small windows around the dial next to the index batons, with the current month highlighted in red. If Rolex releases a watch based on this new patent, it would have a more traditional calendar layout with a central date hand, and peripheral month and day discs.
Rolex has never made a perpetual calendar in their modern era, and for the most part have focused on tool watches with practical complications. Perpetual calendars have been the domain of Patek, VC, AP and other more traditional haute horology brands.
This patent would suggest Rolex is at least exploring more complex calendar functions. Whether they actually release a perpetual calendar depends on strategic decisions about brand positioning. A Rolex perpetual calendar would signal something like “Hey, we can do high complications too.”
Essentially, it would position Rolex a bit differently; not just tool watches, but also fine watchmaking… and their gradual shift to open casebacks would suggest this is indeed in their long-term strategic roadmap. I’d imagine this would debut in the new 1908 line too, not a sports watch.
Technically, I don’t think anyone doubts that Rolex has the capability. They have insanely sophisticated in-house movement development, advanced manufacturing, and decades of experience with reliable mechanical systems. If they want to make a perpetual calendar, they can, and will do so.
But do they want to? I think if this patent leads to a production watch, they will do the annual calendar version first. It’s less complex, requires fewer parts, and fits Rolex’s practical ethos better. It also provides a foundation to to some bug-testing before potentially moving to perpetual calendar later on.
A perpetual calendar would be more of a statement piece, and for the immediate future, it feels like a different message from what Rolex would typically be expected to send.
This patent was filed in April 2025, and published this month. Given Rolex typically works on product development for three to five years before release, if this patent is more “active development work” and not “defensive filing”, we might see a watch based on this in maybe 2028-2030. However, many will already know, patents don’t always lead to products!
So looking ahead, if Rolex is serious about releasing a calendar complication based on this patent, there will be breadcrumbs like additional patent filings refining the mechanism or covering user interface details, movements appearing in certification databases (COSC, Swiss Federal Institute of Metrology), case and dial design registrations showing new dial layouts, and of course everyone’s favourite, trademark applications for potential model names!
As always, time will tell. 🙄
I didn’t plan this post, and I don’t know whether you folks care to read about such technical stuff anyway… I would usually just include a high-level summary of the patent within an edition of SDC-Weekly, but thought I’d try something new.
What do you think?
If you want to look into more technical detail, let’s review some more patent document diagrams!
Figure 6 shows the calendar just before midnight. The roller (43) sits at the peak of cam (33). The energy accumulator is fully loaded and ready to release. Figure 7 captures the first phase of the midnight jump where the roller has dropped into cam valley (33b), releasing stored energy. Fingers 31 and 32 have snapped forward counterclockwise. Finger 32 has contacted lever tooth (12a) and pushed it. Look at where pin (12b) sits – it’s against a radial projection (22b) of the month cam. The cam blocks the lever from retracting.
Figure 8 gives you a detailed view of this whole interaction, where pin 12b is stopped against projection 22b. So when finger 32 pushes tooth 12a, the force transmits through the rigid lever into the date wheel, and the date wheel rotates forward by one step from 30 towards 31.
Figure 9 shows the second phase where finger 31 now drives one of the date wheel’s 31 teeth (1a), rotating the wheel by another step – from 31 to 1. Simultaneously, the intermediate wheel (6) has rotated enough that finger 6b advances the month pinion by one tooth and then the month changes. Both the date and month jump instantaneously, and the patent specifies this happens in about one-hundredth to one-tenth of a second.
Figure 10 then shows the calendar just after midnight. The roller (43) sits in cam valley (33b). Finger 6b remains positioned in the month pinion’s teeth as a physical stop, thus preventing accidental additional jumping. This allows the month jumper spring (8) to be lighter than it would otherwise need to be.
Figure 11 shows the calendar transitioning from say, 30 May to 31 May (or March for example). The key difference from April or other 30-day months is that pin 12b now sits against a circular portion (22a) of the month cam instead of a projection. Figure 12 shows this specific interaction, where the cam’s circular section allows pin 12b to move freely. When finger 32 pushes tooth 12a, the lever simply retracts by pivoting about axis A12; the date wheel stays stationary.
Figure 13 shows the continuation where, after the lever retracts out of the way, finger 31 drives the date wheel forward by one normal step from 30 to 31. The month doesn’t advance because we’re not at the end of the month yet. As the date wheel advances each day, the intermediate wheel (6) rotates because its 31 teeth (6a) mesh with the date wheel’s 31 teeth (1a). Every time the date advances by one day, wheel 6 rotates by 1/31 of a revolution.
After 31 days, wheel 6 has completed one full revolution. Mounted on wheel 6 is finger 6b, which engages with the month pinion’s teeth (2a) once per revolution. So at the end of each month (whether 30 or 31 days) finger 6b advances the month wheel by one step. The month wheel has 12 positions, so it will complete one revolution per year.
Also, if you look at Figure 10, you can see finger 6b positioned between two teeth of the month pinion after advancing it. The finger’s surface (61b) forms a physical stop against surface (21a) of the next tooth.
This prevents the month wheel from accidentally jumping a second time. The patent states this allows the month jumper spring (8) to produce about half the holding torque compared to the date jumper spring (7).
Lower jumper torque means less energy consumption, which means smaller amplitude variations in the balance wheel. For consistent timekeeping whilst operating a calendar complication, this does matter.
Different jumps would be: Normal 31-to-1 advance, 30-to-1 advance, 28-to-1 advance for February in common years, 29-to-1 advance for February in leap years