CHASING THE PERPENDICULAR

 More tracery. This time on St Mary Fratton. Perpendicular style using the tudor arch. Once again there is no way this is going to be parametric. In the vertical direction maybe. Just stretch the lower part which is all parallel lines. But changing the width is going to be complicated. In any case the different sizes of window have different tracery patterns, so it’s going to be a separate family for each window type and the size will be fixed.  First off is a drafting view where I explore the geometry and how much I can simplify for current purposes.

 

 

Then I work up the actual family. Could be a short sweep like last time, but an extrusion would also work because the size is frozen.  I’ve simplified it further on translation to solid geometry. We can always come back and elaborate.  In real life it’s not just and extruded shape of course. The cross section of the mullions has a subtle tapering and curvature, but this is a massing model so let’s keep it indicative. Just convey the impression of gothic tracery.

I arrayed this window type for the full length of the clerestory, both sides.  Definitely an improvement, so let’s keep moving forward.

 

There is a window high up on the sides of the chancel. This can be adapted from the clerestory type. I haven’t quite got the proportions right, but let’s not get bogged down right now. We’re talking broad brush here.

Finally there is a wider window along the aisles.  This one bothers me a bit. Somehow it’s lost the delicacy of the original, but I can’t see a quick fix at the moment. So once again I will be moving on because it’s a big church and there are other window types to tackle including the huge window at the East end. I think I should also do something about the battlements

 

 

TRACING ST LAURENCE

 

It’s time to look at tracery.  Let’s start with another drafting view and a real-life example from Blomfield’s churches.  It’s still a massing model so I don’t want to get too deep into the weeds. St Laurence, West Woodhay, up there close to Hungerford in the NW corner of my study area. It’s a very cute little village church, flint with Bath stone dressings. Typically for Blomfield most of the windows are Early English style lancets with no tracery.  So faithful to the spirit of Gothic and realistically cost effective for a small Victorian church.  Britannia was a wealthy society, but villages didn’t have unlimited budgets. My sketch aims to capture the principles but not necessarily the proportions. It's easy enough to stretch the sill down. I may well be using this basic configuration several times on different churches beyond the world of Arthur Blomfield.

 

St Mary, Sheet is almost identical to West Woodhay in it’s overall massing. The spire is different and the tracery more elaborate. Possibly the budget was more generous. It’s even more obvious that this one is not going to be parametric. Once again you can drop the sill easily enough, but probably be opening up in Family Editor rather than trying to get the tracery geometry to respond predictably to the Height parameter. Let’s keep things simple here. I’m perfectly happy to take a copy of a nested family and adapt it “by hand” to suite another context if that is less effort than making it super parametric so that it adapts to every possible use case simply be typing into the Properties dialogue. It comes down to a judgement call, which is not too hard to make when the geometry reaches this level of complexity.

 

Trying out tracery on an actual church model. All the windows are face-based families. I brought a couple of reference images into plan view for guidance. It may seem perverse to model the tracery as a very short sweep. Basically it’s a way of freezing the shape so it doesn’t distort when the width and height parameters change. Unlike extrusions.

The different sizes of windows, doors and louvred are all types of the same family. Only one type needs tracery. That’s why one fixed size of tracery geometry will do the job. All we need is a visibility parameter to hide it in all the other types.

Reload into the host family. Orbit around the model selecting the types one by one and switching off the tracery. You can see from this view of the East end that the sweep geometry is stable.

 

Inevitably when reloading the family into the project, I start to notice minor issues with proportions, roof slope etc. So there are several other enhancements here. In fact there is another window in the centre of the West End that has tracery. I chose to have two sweeps in the model with slightly different profiles to suit the dimensions of the family type. Both have visibility controls, so they are switched of selectively for the various types. 

This church now has a higher level of detail than I originally imagined, much more than most of the other models in my study. Will I go back and lift the level of all the others? I don’t know, let’s take it one step at a time, and trust my instincts. This is supposed to be a fun project for my retirement from Godwin Austen Johnson. Real buildings with a deep sense of history all within striking distance of the retirement flat which will be my primary residence from April 2026.

 

POINTY VARIANTS

 

I started to look at the original group of church models (not by Blomfield) and saw several that needed another pass.  These two are in the top left corner of the group, a conventional place to start for the Western mind.  I added timber detail to both bell chambers and the one porch.  Fascination how a light touch here and there can bring the character of a model to life. These are both ancient churches, that is to say several hundred years old, with features dating to various times within that lifespan.

Quite probably they both began as single cell structures with no bell chamber, porch or chancel extensions. Saxon or Norman in style. That was my assumption. On checking I find that St Peter, Linkenholt was rebuilt in Victorian times. But with some Norman features reused.  St Mary Mapledurwell is a restoration rather than a rebuild, with features dating back to the 13^th and 15^th centuries.

 

Switch of focus. It’s time to revisit my Gothic arches. I happen to be using face-based families at present but the same principles can be applied to wall-hosted versions. Perhaps I could even nest the face-based families into wall-hosted templates? Might take a performance hit though.

Here is the basic pointed arch with a POINTYness parameter to… make it more or less pointy.

 

Next comes an intermediate shape.  Pointed arch that doesn’t blend into the vertical jambs but meets them at a sharp angle.  This is not very common, but as it happens is used in the church closes to my retirement flat. All Saints, by Temple More, a Victorian church that I haven’t yet started to model.

I added a “Drop” parameter here and used Pythagoras to derive the Radius of the curve. I decided to hide the Drop parameter from the end user to keep things simple. No doubt you can find flatter and pointy-er versions of this type if you look hard enough but for the LOD100 work I am engaged in at present, I think a fixed proportion is less trouble. Just vary the Height and Width and forget about POINTYness.

 

So that was an intermediate step towards the tudor arch, common in Perpendicular Gothic the late style in England before the advent of Renaissance styles from Italy. Here the geometry is much trickier with two radii and two centre points. Once again I have chosen to freeze the proportions of the arch to minimise the chances of the family breaking. 

The hypothetical meeting point of the primary radius and the vertical jamb is derived from a 45 degree triangle keeping the formula simple. Then we need two offsets to locate the centre of the small radius. (H2 and R2)  H2 is dependent on R2, so we only have to set R2 and in practice I will freeze this as a quarter of the width.

 

These are drafting views that I used to develop my strategy. The setting out and the formulae have been transferred to face-based window/recess families and flexed. They seem to work well enough but there will probably be further evolution when I apply them to real churches.