As work began on reassembling the porch balustrade, we realized there were pieces from two different jigsaw puzzles that we were trying to fit together.
In puzzle box #1 was the design of the porch we had demolished a few months before. There were six balusters on the front (east side, on the left in the photo), and four on the south. At the corner, the top railings joined at a 45° angle. But nothing matched; the east top railing was clearly original -- thick, old redwood. Above the four balusters on the south were two boards glued together to match that thickness (once we took it apart, we discovered that there was also plywood used to shim it up to matching height). And finally, there was a long wall, where there was no effort to match the top railing with the height, color, or style of the wood elsewhere -- see detail inset photo.
My original plan was this: I would restore 13 balusters (see part II), the bottom railings, and that top railing on the east. Since none of the top railings on the south were original -- or even salvageable -- I asked Berry's Sawmill in Cazadero to mill a plank 18 feet long (and weighing over 250 pounds). This, I thought, would give me one loooonnng interrupted top railing of new redwood for the entire south side. Over the corner, it would join at 45° with the (restored) top railing. Under that I would attach seven balusters and calculate the width needed for a wall long enough to meet the posts at the appropriate point.
Ah, the best laid plans...
As before, it was architect and preservation expert Mark Parry who set me wise. He first pointed out that the blueprints showed 12 balusters on the south side, not the 7 (or so) I was planning to use. To me, the blueprints couldn't be trusted on details like this; the blueprints also showed ten balusters on another section, but there were nine instead. There were also examples where windows became doors, exterior elevations didn't match the floor plans, and so on. Surely Brainerd Jones was just doodling in enough balusters to fill the picture, I thought. And besides, it was moot; two of the pudgy little columns were damaged, so we didn't even have enough to put 12 on that side.
Contractor David Jessen also began raising questions about the bottom railing. What was I planning to do on the south side? The existing bottom rail was only long enough for six balusters. (See part I -- although four columns were visible, two more were found hidden inside the corner!) Did I have any more century-old redwood rails laying around the place? Exactly how big did I want the shingled corner box, and how would the bottom rails attach to it?
With the pile of unresolved issues growing, Mark, David, and I met in early October to discuss options. We magnified the online copy of the blueprints to highest resolution, then unrolled the actual original paper. Mark pointed out details that I had completely missed -- that there was a cap on the corner box, and a matching reveal on the wall section. By no means was the top railing supposed to be one long slab of wood.
Still the cynic, I. Was there any evidence that the railing was actually built like that? There was no square cap on the corner that recently existed, which would've meant that the top railings were once joined to the corner flat, and not at a 45° angle. And besides (I continued to argue), all of the top railings, including the badly-made ones that were thrown away, were exactly 11 ¼ inches wide -- there was no evidence that wider pieces once capped the corner and the wall section. Then David's recent question came back to me: Did I have any more century-old redwood laying around the place? Why, yes -- yes, I did, in fact.
In the corner of the garage where we found the extra balusters there was also an unusual thick plank of old redwood leaning against the wall. At 80 inches long and 13 ¾ inches wide, its dimensions didn't match anything else around the house. Because one end was cut at 45° angle, I wondered if it was the original south top railing that met over the corner. But there was no sign that balusters were ever connected to its underside, so the original purpose remained a mystery. Now it came to me that I was looking at it upside-down -- if the pointy-end faced west instead of east, it would probably match the angle where the porch meets the house. The three of us carted the board from the garage and laid it on the porch floor, its angle towards the house wall. It fit perfectly. Here was the oversized cap for the south porch wall. The blueprints were indeed accurate, as architect Mark Parry predicted.
We now saw clearly how the puzzle was supposed to fit together. And sadly, we decided that it was best that new top and bottom railings be constructed for both sides, as well as the missing corner cap. Back to Berry's Sawmill for more wood.
(Below: constructing the new wall from pressure-treated plywood, as well as the new corner support)
 |  |
A ripple of other plan changes followed. If we were now accepting the blueprints as gospel, then all seventeen balusters needed to be pressed into service, not just the 13 I was expecting to use. Two that I had deemed unusable because of damage and had been using for glue and preservative tests now had to be restripped with hustle. David would create new end-blocks for the balusters with the worst knocked-off corners, as he pointed out that visitors would see the damaged sections more than any color difference of new wood. Philosophically I was uneasy that we were slipping away from preservation and into restoration, but I know David's advice was absolutely the best, in this situation. He also created a corner cap of appropriate size by gluing together two pieces of otherwise construction-grade redwood that is absolutely remarkable craftsmanship.
Except for repainting the damn floor, the porch saga is over. We considered an awning over this corner, but can't see a way to install one over these irregular angles. We'll watch what happens with the rains this winter, and might revisit the issue. But for now, the work is completed, and decades of mistakes undone; once again you can see a view of Comstock House that looks exactly as it did in 1905.
 |  |
 |  |
Porch floor painting is not a usual restoration topic (nor very interesting), so here's the executive summary: We did it, we were happy with it, then Really Bad Things happened to it.
First, the pre-prep: Scrape, scrape, scrape. All old paint that wasn't firmly attached had to be removed, and damaged wood should be cleaned up for repair. The tools needed are a metal (or firm plastic) scraper, a heat gun, a vacuum cleaner (or blower), and patience. Lots.
(At right: Candice using the vacuum to pick up paint chips)
None of the porch flooring dates back to the 1904-1905 construction of Comstock House, so keeping the existing wood wasn't a preservation effort; we could have ripped it all up to lay down new 1x4 tongue and groove Douglas Fir, as we did with the southeast corner. But except for two spots badly needing repair, the existing floor appeared in fine shape so there was no need for the extra work (not to mention, expense) of replacing it all. We wondered if that was the best decision, however, once work began and found that these sections had already been repaired once before, and with poor workmanship that made some of the problems worse. This meant that our fixes were probably going to be the last these poor old boards could take, so it was especially important to get the job right the first time.
A heat gun was the weapon of choice for this project because chemical stripper could seep into the gaps between boards and be impossible to completely clean up. Heat guns are wonderful tools, but caution can't be emphasized enough; besides the risk of burning yourself (or igniting your house), there's no telling what toxic molecules could be in the miasma of stink that wafts up from old paint melting under the focused heat. Always use a respirator and treat the heat gun with the care you would use with any other power tool -- just because it doesn't have sharp spinning blades doesn't mean it won't kill or incapacitate you.
The oil-base paint scraped up easily after it was softened with heat, but in the larger gaps between boards we found putty earlier had been used to level the floor before painting. Much of that putty had worked loose around the edges, allowing water to pool in these little valleys once the paint was chipped off. Putty is infamous for breaking free any time wood shifts position even slightly, much less when it's at the top of the front steps with constant foot traffic and vibration from the street.
With the cracked paint and putty wedges removed, we were left with a sorry-looking floor. My first thought was to fill those big gaps between the boards with a modern silicon sealer that would remain flexible. But the surface of these damaged areas still had to be leveled, as well as needing "buttered" transitions between old paint and scraped wood. Ultimately I filled the gaps AND leveled the floor using Restor-It epoxy. I'll write more about this excellent product in the future; we're now using it by the gallon on wood, and even concrete, repair.
Below is a before-and-after comparison of the damaged area in front of the stairs.
 |  |
At left below shows the porch after epoxy and with the old paint scuffed up with an orbital sander for better adhesion. The sander was also used to feather transitions between the old paint and wood or epoxy. On the right is the porch primed with Zinser's Bullseye 1-2-3. It will never look this clean again.
 |  |
Once the primer was down, the porch was blocked off for two weeks. Four days drying for the primer, then two coats of paint (Kelly-Moore French Sonnet KM3490-1 gloss, for my own future ref). It looked terrific. It looked REALLY terrific.
But about six weeks later, I noticed a small bubble on the southeast corner during the peak heat of the day. Then another, larger this time, also in mid-afternoon. And another. By the time contractor David Jessen returned to complete work on the railing, paint and primer were easily lifted up from the ends of the boards. "It's like picking at scabs," he mused laconically, "if you like that sort of thing," peeling off another fist-sized sheet.
I deeply feared that soon the whole porch would soon erupt into a mass of similar boils. The mailman would kick up clouds of off-white paint on his way to the mailbox; the cat would forge a smaller trail of exposed wood on her route between the front door and the sunny corner.
Everyone had a different explanation for why this was happening. The Benjamin Moore dealer who sold me the primer blamed the paint; the Kelly-Moore dealer who sold me the paint said no primer should've been used. (Can anyone explain why are all paint stores must be named "Moore?") I also asked the knowledgeable paint people at Friedman's Hardware where I had purchased the preservative used to treat the floor boards five months earlier; none of them had heard of such a problem, but implied that it was probably was my fault for stupidly not buying their brand of porch floor paint.
I experimented and observed. Was it the otherwise-reliable Zinser primer that failed? Following Kelly-Moore directions, I diluted their paint by 20 percent as a primer, and painted that over a peeled-up section. A few weeks later, that wasn't sticking to the wood either. I also found that there was no sign of any problems whatsoever on the old flooring -- the only paint coming up was on the porch corner where there was new wood. The primary suspect was now the two good coats of Green's Clear wood preservative that I had applied to the boards before the floor was installed.
Reading again the can of Green's Clear, the directions stated, "PAINTABLE WITH OIL BASED PAINT WITHIN 48 HOURS" (emphasis mine). Would that I could -- alkyd-based exterior paint is no longer legally sold in California. Zinser 1-2-3, however, boasts that it's "great for hard-to-stick surfaces" of all kinds. Could it stick to wood treated with Green's Clear? Apparently not.
I e-mailed Green Products Co. and explained my problem. Their response was that I should strip the paint and primer, then scrub the wood with TSP, followed by a solvent-based primer -- in other words, remove as much of their product as possible from the surface.
There's rich irony that Green Products, manufactured in Richmond, California, makes a product that really can't be used as directed in the same state. At the very least, the company should have warned buyers that their wood preservatives absolutely cannot be used with the allowed latex-based products. As I will spend the first rain-free weeks of summer scraping up the paint and primer on the corner of the porch, then scrubbing the surface down with TSP, then coating it with Preserva-wood, then more fresh paint, you can bet I'll dwell long and hard on how I'll never use a product of theirs again.
With the porch floor completed, the next big job was restoring the balusters (the little columns in the porch railing). I expected the job to take six weeks, two months at most; but as the calendar drifted into the tenth week, I began to realize that I had greatly underestimated the project. It ultimately took over five months to prepare the posts for installation.
My first mistake was jumping into the project without a clear idea of what actually was to be done, a misstep dissected in part four of this series. Regarding the balusters, I didn't even know how many I had; on the porch that was just demolished, there were 6 on the east corner and 4 on the south side. Another five were in the garage, still attached to part of the sawed-off railing. That added up to fifteen -- but I planned to restore only 13 of them because two had badly damaged corners. Then during demolition, two more balusters turned up hidden inside the corner post (see part I). Now our little family had grown to 17.
The difficulty in removing the paint was also an unexpected problem. I knew that serious work would be required; in the past I've stripped furniture, doors, and windows using both a heat gun and chemical goo. The latter is messy and dangerous, requiring a respirator, rubber gloves, and goggles -- an uncomfortable way to sweat off a few pounds on a hot summer day -- but it's what you have to do if the wood beneath is irreplaceable. As Candice says, "stripping is an act of love." (Methinks our Google search hits just jumped a thousandfold...)
To strip the balusters, I used a plastic mortar mixing tray -- essentially, a big cat litterbox. Because the balusters are narrower in the middle, I wrapped two paper towel rolls in heavy aluminum foil, then covered the tray and rolls with more foil. (Two sheets had to be crimped together to make the foil wide enough.) This gave me a perfect form to hold a baluster. Over that I poured enough Jalisco stripper to cover about one side of the wood and let it cook away for about 40 minutes, turning it over at the midpoint to coat the other half.
To remove the paint, use a putty knife with a thin, semi-flexible aluminum blade (plastic knives will start to melt after a few minutes). Gently scrape off the stripper/paint gunk, wiping the residue on the blade with newspaper. Try not to let the stuff fall back into the tray -- even though the stripper gel is now the same color as the paint, you can store it for reuse if it's clean enough. Typically half of my "pour" was stripper from the last batch.
After scraping, I used a stiff wire brush over the curvy details at the top and bottom of each column. Then the entire piece was scrubbed down with heavily-diluted paint thinner using coarse steel wool.
All that work removed only about half the paint; repeat everything the next day. A few balusters required a third trip to the dip.
Coarse sanding followed, always using #60 grit. I used an orbital sander for the cylindrical columns, beltless sander for the flat surfaces, and those wonderful 3M sanding sponges on the details. A final rubdown with a well-worn sanding sponge finished it up.
Last came patching cracks and holes from pulled nails. An exterior-grade wood putty or glue was needed, but it was important that the color of it blend perfectly. Earlier in the year, I had experimented with mixing vintage redwood sawdust with four products: Durham's "Rock Hard" Water Putty, ZAR wood patch (neutral color), Elmer's Wood Filler, and Elmer's Stainable Wood Glue. After thoroughly dried, I applied a coat of linseed oil, followed by the turpentine-linseed mix that would be used on the balusters. All products worked reasonably well. I chose to use Elmer's Stainable Wood Glue, mixing it with sawdust until it has the consistency of toothpaste. Cracks were filled with the mix on tips of toothpicks, larger holes smeared up with Popsicle sticks. (I swear, more repair work on this large house is done under a magnifying glass than I would ever have believed.) After thoroughly drying, the glued area was sanded again. Although the patch appeared darker on the unfinished wood, it was almost invisible once the piece was oiled.
The total time required: About 40 minutes for each session in the dip, then another 40 minutes of labor; 40-50 minutes of sanding, and then often another 20 minutes of patch and resanding. Figure on 3+ hours work, spread over five days. For every baluster.
The lastest step was dousing each column in linseed oil (because this is old wood, I applied two coats) before it was installed on the railing. Below is a before-and-after picture of the girls.
 |  |
A footnote, regarding Comstock House archeology: The balusters that were hidden in the column or in the garage had a coat of lead paint in a redwood-y color, and the 10 that were on the demolished railing had an additional coat of sickly mocha-colored latex. But in his specifications, architect Brainerd Jones had insisted that the exterior wood should never be painted. Was there any evidence that the balusters were originally just oiled, as he specified?
Although that old lead paint stuck like glue, there were a very few sections where a small section could be lifted off intact. And sure enough, that century-old heart redwood underneath had just about the same rich color as predicted from tests of the turpentine-oil treatment.
 |  |