Timber Frame Water Mill Reconstruction Project, UK Part 2 – Framing the Roof

Written by Guest Author, Gavin Sollars, Timber Frame Carpenter, UK

The proper proportioning of roofs forms one of the, if not the most important branch of the art of carpentry, testing alike the manual dexterity of the craftsman, and the taste of the designer.

George Ellis, Modern Practical Carpentry, 1910


In Part 1 of this three part series about the reconstruction of an historically significant timber frame water mill located on the beautiful River Test in Southern England for which he was responsible, Gavin shared some background about the project, some design details, the challenges his team faced at the jobsite, and the fabrication, delivery, and assembly of the green oak timber frame on-site. In this post he will share some details about framing the roof structure. If, Gentle Reader, you have not read Part 1, you may want to do so before reading further.


Framing roofs has always been the most satisfying aspect of carpentry and one to which I have dedicated the most amount of study. When I first began learning the trade I was fascinated by the way more experienced carpenters succeeded in making differing planes and angles intersect. I often struggled to get my head around how they made it all come together. From those early days I made it my aim to soak up as much knowledge as I could from other carpenters about their approach, how they overcame problems and most crucially the ways in which they worked. 

During the time I spent with the Compagnons Du Devoir in France I was introduced to a way of thinking about the roof as more than a functional structural component, but rather the highest expression of the carpenters’ craft, indeed timeless art.

Compagnons Du Devoir has regulations, and one of my favorites reads as follows:

“Individuals are invited to sow beauty with their hands, hearts and minds.”

Compagnons Du Devoir

One only needs to see some of the chef d’oeuvre (masterpieces) that Compagnons Charpentiers have completed, many of which required in excess of 1,000 hours to complete, to understand the deep respect these craftsmen have for their trade.

In addition to my time with Compagnons Du Devoir I have been very lucky in my career to have worked under and alongside many extremely talented carpenters and craftspeople who generously shared their time, knowledge and skills with me. As the years go by I have come to a greater appreciation of the fact that the information and techniques they passed on to me were in turn passed on to them by other generous craftsmen in the past, and so on down through the ages. The knowledge we have today of structural woodworking is a gift from many generations of carpenters who worked to perfect their craft, serving their communities while at the same time training future generations.  

While the roof is essentially about providing shelter from the elements, one of the most basic human needs, over uncountable millennia carpenters the world over have built diverse roof structures for diverse conditions, to perform in different ways and to convey many meanings far beyond simply keeping the rain off – some as a display of wealth and power, others as a show of skill and mastery. Many stunning examples are breath-taking monuments to the earthly representations of the deities they protect.

The stunning wooden Muqarnas vaulting that forms the ceiling at Cappella Palatina, Palermo (Palatine Chapel), built by the Normans (completed 1143 AD). Having conquered the region, they fused the craftsmanship of the Byzantine, Norman and Fatimid traditions- A spectacular display of power, cultural understanding and dedication to God. 
The Pantheon, completed in 126AD Rome, Italy. Originally a temple, later a church and even government offices, now a tourist attraction. A tremendous feat of roof engineering in any age, one every carpenter and architect should visit.
The Pantheon at night
The Pantheon’s coffered domed ceiling, also the underside of the roof structure.
Filippo Brunelleschi’s Duomo at the Santa Maria del Fiore cathedral in Florence Italy, completed in 1461. 45 meters diameter. A marvel of both roof engineering and timber framing, it was the largest dome in the world for several centuries. In addition, the drafting techniques Brunelleschi invented to design it remain the basis for all architectural and engineering drawing even today.
The ceiling of the Duomo. The little hole above the soldier’s spearpoint is a window between the inner and outer domes visitors can peer through.
Horyuji Temple, the oldest surviving wooden structure in the world. All wooden construction
The 5-storey pagoda at Horyuji Temple, Nara, Japan is 32.5 meters in height (106 feet) and is one of the oldest extant wooden buildings in the world. All wooden construction
The pagoda at Yakushiji Temple, Nara, Japan. Built in 680 AD, this all wooden structure is listed as one of Japan’s National Treasures. Clearly, for this and the other buildings pictured above, their roof structures define, beautify and give spiritual meaning to the building beyond just keeping the rain off.

What I aim to show in this article is how we framed out the roof for this humble watermill project, in particular the two valleys – areas of the project I was directly involved in. Some of the elements are complicated so I have included photographs and drawings to aid in visualizing.


The reconstruction of this watermill was undertaken in two phases. The majority of the main house (with one side along the river Test) and the original watermill structure were destroyed by fire in 2018. In the aftermath of the fire the main house building had been repaired by a contractor, leaving us the bare bones of the original mill on the other side of the river to reconstruct. The main contractor had left us an exposed gable end on the already restored house to connect our frame to the existing dwelling. This was achieved via a small link building at right angles to the mill and ultimately spanning the river. 


We cut as many components as possible in the workshop, either from drawings or by standing the roof components up in the shop and working directly from them. In the long run this way reduces expensive site assembly time and it’s generally easier to complete the work in the comfort of the workshop protected from the English weather. On this occasion however, it wasn’t practical to pre-cut everything in the workshop because of the many unknowns, the impending assembly date, and the high risk of critical components not fitting correctly at the job site.

Ultimately the two buildings did not end up square to each other creating a sort of crushed box geometry effect in the roof that joined the structures. While not a problem in itself, this unusual geometry complicated matters a bit. Small variations made what should be even and regular roofing lengths and bevels suddenly slightly irregular, amplifying small discrepancies over distance. 

My team cut the simple common rafter pairs in the workshop. They were joined at the apex using a pegged bridle joint (see sketch below) with the seat cuts pre-cut based on measurements taken from our drawings. We also pre-cut the bridle connections at rafters that either met a valley, or formed an opening for a rooflight or dormer, but left long them long and trimmed to final length on-site.

A sketch of a bridled common rafter. Essentially an open ended tenon and mortice, this joint performs well in compression as the weight of the roof on each side pushes the two parts together. Both the tenon and mortice are one third the width of the timber and pegged together.
A seat cut on a common rafter.

With the structural frame assembled, two members of the team set to work fitting, pegging, and nailing in place the standard common rafter pairs whilst I and another worked on the purlin returns and the valleys. When the framing of these elements were completed we were joined by another colleague, Jamie, who framed out the hipped gable end, dormer and rooflights.

Purlin Returns   

Definition of a Purlin – “Horizontal beams supported by the trusses between the ridge and the wall plate that carry the common rafters” Corkhill, T., 1979. A Glossary Of Wood. London: Stobart Davies, pp.431,432.

In the run-up to this job I held an interim leadership role – this watermill was one of the first jobs I had overall responsibility for, and to date one of the largest roofs I have worked on. From the instant I first saw the drawings I realized these purlin returns would be one of the more difficult elements. 

One of the more unusual things of note about the way in which the purlins are framed is that they are clasped between the underside of the principal rafter and a short tie. This method was often traditionally applied on trusses with smaller sections where the size of the principal rafter would decrease above the purlin. However in this case we cut a scallop (seen in the rendering above) to allow the purlin to be rolled into it’s housing after the trusses were in position.

The purlins are positioned at the same elevation around the whole building, which means that at the intersection of the link and the main frame one returns into the other and wraps around a principal rafter, throwing up the slightly odd compound cut shown in the rendering above. Ordinarily purlin returns can be tricky enough to get right, they often result in either a mitered cut or one notching around the other. 

With the help of our draftsman and my roofing square I made a test piece to take to site to aid in tweaking the final fit where necessary.  

On site after some careful measuring, test fitting and a little adjustment we got the returns installed.

Valley Rafters

The next step in the process was to pitch the valley rafters. On each side there was a lower and an upper rafter. The lower one was relatively straight forward, springing from the top plate (or wall plate) and striking the side of the principal rafter. However the upper sections were a different kettle of fish.

Here you can see the upper and lower valley rafters in place.

Where the valley struck the purlin a complex cut wrapping around the top arris of the purlin was necessary (shown below) before striking onto the side of the principle rafter. This took a little bit of trial and error, but with my colleague Dom’s assistance in figuring out a couple of the bevels, we got them cut and fitted with satisfying results.

Jack Rafters 

Definition of a Plane – “A flat surface; one in which any two points lying on the surface may be joined by a straight line lying on the surface” Corkhill, T., 1979. A Glossary Of Wood. London: Stobart Davies Ltd, p.411.

Definition of a Layboard – A layboard is a board of timber fixed to the rafters of one pitched roof to take the feet of the jack rafters of an adjoining roof.

Once all the head scratching over complexities was out of the way it was onto cutting the jack rafters to length. You may notice that the above photograph of the two valleys shows them sitting in the plane of the main roof similar to a ‘layboard.’

A design like this has few advantages for cutting the feet of jack rafters. On one side of the roof the cuts are beveled across the face and square on edge making them simple to cut. And on the adjoining roof there is again a beveled cut on the face but with a seat cut angle on edge. Whilst one carpenter trimmed out for the rooflight window, two others set about cutting and fitting the jacks to the left of the valley. As this was progressing I concentrated on determining the lengths for the right hand side and cutting the pairs on the adjoining link building. Once cut, the rafter pairs were raised one by one and nailed off. We used galvanized wire nails where they would be concealed from view (bright steel corrodes badly with the high tannin content in green oak). And where visible we used tapered rosehead nails for a more decorative finish.     

Once completed we stood back and admired the work. Everyone involved put in a great deal of time, care and effort to ensure this frame was both structurally sound and looked the part. Leaving nothing behind but a hand-jointed timber framed building of this sort of size and quality was very satisfying.

I hope my Gentle Readers have gained some insight into the basics of how traditional timber-framed structures like this are built, and how, despite using more modern techniques to do the “grunt work,” the ways in which these buildings are constructed has remained fairly unchanged throughout the generations. 

Gavin Sollars

If you have questions or would like to learn more about our tools, please use the questions form located immediately below. Please share your insights and comments with everyone in the form located further below labeled “Post Comment.” We aren’t evil Google, incompetent facebook, or sneaky data miners and so won’t sell, share, or profitably “misplace” your information. No sir.

Other Posts in Timber Frame Water Mill Reconstruction Project Series


Timber Frame Water Mill Reconstruction Project, UK Part 1

Written by Guest Author, Gavin Sollars, Timber Frame Carpenter, UK

The village is the token and pride of England; there are usually found in it vestiges of earlier life – cottages, manor-houses, farm-houses, with buildings of more or less historic interest; and who should understand them, their origin, their peculiarity of structure, better than the local carpenter?

Walter Rose, The Village Carpenter 
The reconstruction job site as seen from across the River Test. Notice the green netting installed to prevent things from falling into the environmentally-sensitive river.


I became acquainted with Stan early in 2019 when I decided to look at buying an Ootsuki Nomi. During my search I became skeptical of many of the Japanese tools that are widely available to the European market and, after a lot of research, I came across Stan’s contact details and sent him an email. Stan took a great deal of time to discuss with me what really motivated my purchase, the kinds of things I should take into consideration when looking at Japanese tools and went into detail about the intricacies of Japanese craftsmanship. The information he freely provided was invaluable, and with his help I feel I made a very good choice, and now have a tool that will serve me for my entire career.

I recently updated Stan with some pictures of buildings I had worked on and he asked if I would be willing to share them with the readers of his blog. The overall aim of these ramblings is to describe to you (who Stan calls his “Gentle Readers”) the general outlines of the reconstruction of an 18th century traditionally-jointed timber frame structure in a beautiful area of the English countryside in the summer of 2019. I hope that this article will give you an understanding of the work that was undertaken and also the enthusiasm I have for this archaic variety of building craft and carpentry as a whole. 

Project Background

Timber framing in the UK has enjoyed a resurgence in popularity over the last 30 or so years, with quite a number of small to medium sized specialists in the craft building new ‘post and beam’ style buildings that emulate the traditional frames of the past. These companies are mixing time-honoured design details and timber framing techniques with more modern methods of production: chiefly circular saws and portable chain mortisers to rough out the work. This deep understanding and appreciation of historic building vernacular made my employer (The Green Oak Carpentry Co.) well placed to undertake the reconstruction of this Project.

The Mill prior to 2018 fire

The company was awarded the contract to reconstruct the building as close to the original as possible.

The Project is situated to the North of the Test Valley in Hampshire County in Southern England on the banks of the beautiful River Test, famous for trout fly fishing and “gin-clear waters.” The original waterwheel powered a grain mill. It was later converted to paper production, and even later housed a generator serving the nearby Manor House. Unfortunately, the original structure was completely destroyed by fire in early 2018.

At 21 metres (68.8’) long and 5.7 metres (18.7’) wide, the main structure is situated on a small island created by a man-made diversion in the River Test. The river flows from the north and is then diverted via a sluice gate to the right. The river then widens into a pool and bubbles quickly along the west side of the building. The diversion along the east flank drives the turbine, and passes underneath a wing that links the main structure back to the existing dwelling and also houses the turbine and mill workings. 

Historical & Structural Considerations

The original building was “listed” with the Historic Buildings and Monuments Commission for England confirming the historical importance of the building on the one hand while placing restrictions on how any work to the building can be performed on the other. What remained of the original building after the fire was still subject to the Commission’s rules and regulations, of course. It once featured shorter posts sitting on top of a brick wall from approximately first floor height. In the wake of the fire the surviving outer brick walls were deemed too structurally unsound to bear a load – however, due to the walls being “protected” under UK law, they had to be preserved. To get around this issue we installed two outer plates running around the perimeter of the building with the lower of the two (D – below drawing) supported on metal brackets (C) connected to the back of the timber Jowl posts (B) by lag bolts. The full weight of these two plates, as well as the softwood stud wall with conventional insulation and weatherboarding is carried by these brackets transferring the load to the jowl posts (B).

A detail drawing (drawn by myself) of the steel bracket, showing how load was removed from the fragile existing wall. The drawing also explains the interplay between our frame and the other elements in the building.

Design & Construction Details

Framing work started in July 2019 with a team of eight carpenters framing the bulk of the structure over a period of five weeks in workshops off-site. A team of four transported the fabricated components of the timber frame to the jobsite, assembled and raised the frame, framed the hips and valleys, then fitted the common rafters and cut and assembled the jack rafters.   

Constructed entirely from European green oak, the structural frame is very utilitarian by design and lacks the aesthetic details like the curved braces typical in many historic timber structures in the UK. Nonetheless, it has some nice detailing that might not be obvious at first glance. 

The main posts (wooden structural columns) are mostly jowl posts (aka “gunstock posts”) that flare at the top with tenons that fit into both the top plate (beam running along the top of and connecting the exterior posts) the column and tie beam. Historically, jowl posts were cut from the flaring grain of the base of a tree. These butts were often quartered and each post placed in the building adjacent to its sibling. I believe that this is a similarity historic English carpentry shares with its Japanese counterpart.

Here you can see the cross frame construction 

The dominant style of cross frame (or bent in North America) features a bridging beam (the large beam that spans the first floor and carries the common joists), a tie beam which spans the top plates. This beam stops the wall frames from spreading under the load from the roof. And then a simple truss design consisting of two vertical studs and an upper collar with short stub ties jointed horizontally between the principle rafters and studs. The purlins (the members that run the length of the roof) are ‘clasped’ between the stub tie and the principal rafter.  

The main roof frame is comprised of bridled common rafter pairs, a pre-Georgian (prior to 1714) hip gable at one end and a standard gable at the other. A pre-Georgian hip is the English name for a hip rafter that is usually square in section and canted so that one edge is in the plane of either the gable or the main roof. Hip roofs were historically framed in this way until carpentry methods changed and more of a ‘hip board’ set plumb with jacks pitched onto the sides became the preferred method of hip framing. 

The pre-Georgian hip gable. The effect of canting the hip rafter in this way means that the jack rafters pitched from the two walls have a side cut angle on top and a square cut on the edge, but the jack rafters on the gables have a more complex (and comically named!) ‘nuns crutch’ or ‘lip cut’.

The adjoining wing has a wider span and a higher apex to the main building, and the roof meets the main roof in a valley. These valleys are similarly canted into the plane of the roof like the hips. In the same way the hips produce two different jack rafter cuts so does the valley. You’ll notice that on the main building there is a square jack cut and on the linking building the jacks have a compound cut onto the valley rafter. After running into the valleys, two small hip rafters pick up the opposite slope of the main roof. All of these angles were found using the framing square. 

The junction between the wing and the main building looking back down the east wall.

The main building is split into two clear halves; one which is vaulted floor to ceiling, and the other which has two floors of joists.      

Green Oak Timbers

What sets this type of carpentry apart from other woodworking is its use of timbers that are rough-sawn and often of irregular dimensions, requiring an understanding of how to deal with imperfections. For example, timbers are often significantly out of square, and dimensions only approximate: according to the European standard allowable tolerances are +9mm ~ -3mm in section. 5mm of deflection per metre is also allowed. These significant irregularities complicate the carpenter’s job.

Moisture contents can be in excess of 60% in fresher felled stock and during the summer months the warmer weather can cause problems with drying and shrinkage – we often keep our stacks covered with hessian cloth in an effort to shade the timbers and reduce warpage and cracking. 

Timber framing using this challenging material teaches a carpenter much about the nature of wood as a living thing, the characteristics of the timbers, how they are likely to behave and what can be asked of them.  

The Layout Process

Carpenters have developed various layout systems over many centuries to overcome the difficulties of working with irregular timbers. In my company we use lofting, for example. According to this method, we draw out entire layups on the floor and align the outside edges of each member to these lines with any sectional irregularities placed on the inside of the building, whilst any crowning (deflection or bowing) is oriented upwards and outwards.

Once the layout is drawn full-scale on the lofting floor, the timbers are placed on blocks in alignment with their corresponding grid lines on the floor. The various members are then laid one over the other so carpenters can accurately mark out lengths and scribe the shoulders of the joints using levels and plumb bobs. 

The plumb bob is an ancient tool that you are doubtless familiar with – its importance in carpentry cannot be overstated. Plumb bob scribing, or ‘scribe rule’ layout, is a difficult thing to describe without actually seeing it done. What it boils down to is using a perfect reference in a less than perfect situation. By sighting down the plane of the timber by eye and comparing it at the same time with a plumb line a carpenter can gauge to what degree that face is out of plumb and then accurately replicate that plane on the shoulder of the intersecting timber. The shoulders of tenons and widths of mortices are carefully marked using this technique.

Once a frame is cut and fitted back together, a plumb bob can be used to accurately transfer and mark additional details up from the floor. In the case of this watermill I used a plumb bob to mark the theoretical positions of the purlin housings on each truss. The result was that, regardless of the amount of deflection or variation in thickness of each principal rafter, the purlin housing was maintained in a consistent level position down the entire length of the building.

A plumb bob being used to mark the principal rafter to tie beam joint. On the floor you can see the layout lines. We mark out each individual frame in a building full size on the floor with chalk lines, and then set up the timbers to those lines. This plumb bob belonged to my great grandfather who was also a carpenter. 

Frames are generally made up of wall frames (running the length of a building) and cross frames (spanning a building). Many of the timbers are therefore used repeatedly in multiple layups. In the case of main posts they are first framed into the exterior walls of a building. Once the walls are completed they are framed in the ‘cross frame layups’. To ensure that the posts return to the right height and rotation that they were in during their previous layups, we use scratched datums (often a set distance from the top of top plate) and rotation marks that allow us to wedge the timber to return it to plumb or level. 

Ensuring that a designated point on each timber is plumb and level is essential, particularly for those in multiple layups. It guarantees that a timber has been returned to the correct plane each time it is fitted up so that when it is stood up and connected to multiple beams, the rotation of each individual shoulder scribe is correct.   

Once the bulk of the main structure is framed we laid-up floor joist and ‘cogged’ the tie beams into the top plates. Traditionally tie beams were dovetailed into the plates but because the shrinkage of dovetails (green oak, remember) tends to cause the building to spread, it is now more common to see a simple cog used. A cog also slightly outperforms a dovetail in green timber when under tension.

Assembly and Erection at the Job Site

The building enclosed two concrete pads differing in elevation by about 200mm (8”). The base of each post therefore was designed to accommodate this change maintaining the design elevation of the building. This and other variables made laying-out of the building one of the most challenging I have ever been involved in.

This photo shows the “jowl posts” and the boom of the spider crane as assembly work is underway.

The only access to the site was a track through a field at the rear of the building and a small trackway and bridge over the river too narrow for a mobile crane to cross. The solution we employed was to bring in and set up a small spider crane in various places inside the building. The limitations of this machine required us to be very methodical about the assembly sequencing to ensure we didn’t obstruct subsequent lifts. 

Space was at a premium. Without a large area to unload the piles of timber, we had to unload the timbers and other materials in the field behind the property and then use a remote-controlled tracked carrier to ferry timber across the narrow access bridge.  

And, to throw one more element into the mix, the river is an extremely well-protected ‘Site of Specific Scientific Interest’, meaning we had to be very careful when cutting roof members to prevent sawdust from drifting into the water course. The scaffolders installed netting around the entire perimeter to prevent anything from falling from the scaffold. We also did the majority of our cutting away from the edge of the scaffold on a plywood deck we laid on the joists.

As the building began to take shape its scale became more apparent. At nearly nine meters high, it’s an impressive structure. 

Lessons Learned

I took away a lot of lessons from this building project such as managing levels on-site, and the importance of every trade singing off the same song sheet, as it were.

We had issues with the initial layout of the structure as it became clear that the structural steel that effectively served as the starting point for everything above it had not been installed at the correct elevation. The work was delayed while we sorted out this problem.

I also learned valuable lessons about effective joint placement. Because of the space constraints mentioned above, we were forced to erect the structure by lifting and placing each cross frame and then linking it back with its purlins. However, because the scarf joint landed on the wrong side of each truss, every time we craned a purlin into position, it was left temporarily unsupported at one end. These decisions were admittedly made early on before any proper site visits were made. A proper method statement was in place, of course, but the experience taught me that starting with the end in mind is important when planning.

I hope that you, Gentle Reader, gained some insight into the work that I am involved with and found it an interesting read. If you would like further information about historic timber framing in the UK, I recommend a small book titled Discovering Timber Framed Buildings by Richard Harris. 

– Gavin Sollars

Gavin Sollars hard at work

Thank you for your reading this article. It is rare to find a craftsman like Gavin with the skills and inclination to write about his work and a willingness to share it freely with others. Gavin did not write this to promote his company, but if you like this sort of thing as much as we do, please visit his company’s website and sign up for their newsletter.

In the next post in this series Gavin will outline the construction of the roof frame. Please stay tuned.


If you have questions or would like to learn more about our tools, please use the questions form located immediately below. Please share your insights and comments with everyone in the form located further below labeled “Leave a Reply.” We aren’t evil Google or incompetent facebook and so won’t sell, share, or profitably “misplace” your information. Stick a needle in my eye.


Other Posts in Timber Frame Water Mill Reconstruction Project Series

Part 2 – Framing the Roof