Michael Spencer
Artist Blacksmith

Chancel Screen

St. Mary's Anglican Church
Crousetown, Nova Scotia, Canada

[View of the chancel screen]

This combined communion rail and chancel screen is in a small Anglican church in rural Nova Scotia. The congregation is small but has a number of devoted supporters of whom my client is one. I'd worked with him before on lighting for the same church and the chancel screen came as an exciting challenge.

The space on the platform was limited, especially as it has to accomodate the Baroque ensemble which has presented summer concerts here for twenty years, so the ironwork was to combine screen and rail in one piece. Because the Anglican Church permits a chancel screen only if it doesn't obscure the congregation's view of the altar, I could resist my temptation to fill the whole 21-foot space to a ten-foot height (or even to the ceiling) with iron. The client provided a sketch suggesting the Tudor rose motif, which I had already used in the lighting, and the ogee arches. The rose and the fleur-de-lis are traditional emplems of St. Mary.

To ogee or not to ogee

The traditional way to construct an ogee arch is to draw two tangent circles, one of which passes through the springer point, other through the apex. The line of the arch follows one circle from springer to tangent point, then the other to the apex. I tried a number of these constructions and found them all too stiff and heavy-looking. Some free-hand curves looked alright but enlarging them for the larger center arch so that the eye saw them as somehow "the same" proved difficult. So I tried generating curves from the followng equation, of which the equations for the circle and the ellipse are special cases:

                 Xm      Yn
                ---  +  --- = 1
                 Am      Bn

With a programable pocket calculator and graph paper I could try out different values for "m" and "n", which determine the curvature, and, for the shape I liked, change the "A" and "B" values to the dimensions of the center arch and run it once more to get a shape that was "same" but larger. This would all have been easier with Mathmatica or a CAD progam but all I had at the time was the calculator. Incidentally, this equation is related to Piet Hein's "superellipse" or "superegg'.

New iron in old buildings

It was fortunate that I thought to set up a builder's level and survey the church. The platform looks level but is 1 1/2" higher on the right than on the left. You can see the compensating taper in the flat panels below the roses but better that than to have had the verticals visibly out of plumb. Because each half-arch was a different length, the springer line became the reference level and each piece was center-punched at that point. The curves were formed on a jig from that point upward and, after surface texturing, the legs were cut to a length from that point downward that suited the spot on which it was to stand. For trial set-up in the shop the arches could be shimmed up to allign the center-punched springer line with a level, taught wire stretched at head height.

Doing the details

[Detail of moulding] The idea for the molding bosses came from a piece that I saw in the shop of New Hampshire blacksmith David Court. It proved impossible to get the diamond bosses oriented uniformly with a stake die and a hand-held punch so the stake tool became a fixture. Two "walls" spaced to accomodate the 1/8" x 2" bar when hot, a "roof" with a square hole to take the punch and, over a rear extension of the die plate, the cam operated plunger from a valve spring compresser which would quickly clamp the stock in place, leaving both hands free. In a second step, the punch, point up in the vise, made a stake over which to hammer up sharp the point and edges of the pyramidal bosses which had emerged from the die rather rounded and amorphous-looking.

The D-shaped depressions along the lower edge were made with a similar fixture, but with a spring-retracted punch. Each one required one blow with a 5# hammer, about five per heat. Because they stretched only one edge of the stock, they resulted in pronounced edgebend which was cured by vigorous crosspeening along the back of the opposite edge.

The twisted element of the molding is 1/4" hex bar which gives a softerlooking twist than square stock.

The client had suggested openwork roses but I wanted to try somthing demonstrated at the Towards a New Iron Age conference by British blacksmith Stuart Hill, a "cut out, forge and weld back" technique. The railing was to be only about 24" high because a greater height would be awkward for kneeling communicants and this meant that the square panels that fit within the opening would be small enough to be easily managable in plate. I had the squares sheared and the holes torched by the supplier. Five tear drop shapes cut from each cut-out disc were textured with a 1/4" diameter fullering die under a 25# hammer which spread them enough that, after the edges were curled up, they still more than filled the circular openings. The edges of the 3/16" plate panels are upset to about 1/4". Turning up the edges by hand distorted the plates but produced a softer look than would have resulted from having it done on a box brake. It would have been really nice to have had a gas forge so that the whole plate could have been hot at once when I trying to get them flat again!

[Detail of the rose
panels] The rose panels were glass-bead blasted at the local automotive speed shop. All the other parts were pickled clean in phosphoric acid and everything treated with Oxpho-Blue(tm). Two 750-watt electric paint strippers with the heat shields taken off made good immersion heaters to keep the acid at working temperature in a chilly shop.

The sconces, from 16 ga. and exhaust pipe offcuts, are buffed to a mirror finish and have keyhole slots which fit over and snap down onto panhead screws in the brackets. They can be snapped off for cleaning and be snapped back on again. Using the sconce brackets as spacers between the pillars was a fortunate design choice, because the ends could be ground during final fitup to compensate for unevenness in the pillars and make the whole screen pull up true when the bolts were tightened. And, because they projected beyond the plane of the screen over their full five-foot length, they work effectively to give the screen three-dimensional quality where very litle space was available.

Straightening circumstances

The greatest problem was corkscrewing of the vertical parts of the arches. They were worked all over under a 25# hammer and truing up the die faces reduced but didn't eliminate the problem. The arch/rail sections were fitted up on a rigid jig, so they came out true. But when all five sections were bolted together, everything pulled out of line because of the slight corkscrew in the pillars and it was a lot of probably avoidable work to fix it reheating each arch half and tweaking it true.

[Detail of a rose] I had expected installation to reveal all my oversights and ignorance, squeezed down to the end of the project like the tangles in a rope you're coiling up. But it only took a couple of days including sawing through the floor of the church to get at the crawlspace and all went smoothly. I knew it was looking good when the client volunteered to drive my wife to her nursing shift at night in a snowstorm sooner than interrupt the installation.

Original photos by Peter Barss
Updated 9 May 2000 -- Mike Spencer