When the roof shingles were removed, we hoped to finally solve the last big architectural mystery of Comstock House - but what we found was completely unexpected, and has led to other questions we still can't answer for certain.

The puzzle concerned the rain gutter system in the front of the house. We knew that there were downspouts inside all three of the columns, which were (presumably) attached to the back of the original gutters, and the aesthetic goal probably was to avoid big ol' metal downspouts from spoiling the look of the otherwise all wood-and-window house. Nice, but drawbacks include the risk that heavy rains could overwhelm such an enclosed system and cause leaks inside the house (the architectural history of this topic was discussed in an earlier essay). However that system was designed, we believed it apparently wasn't very successful. Between the earliest known photo (probably 1905, a few months after construction) and the 1908/1909 postcard, a rainstop board appeared above the actual gutter, likely intended to slow the torrent of rain as it shoots down our steep-pitched roof. Even that didn't perform very well; during the heavy rains of 2009-2010, every downpour turned the front of the house into a waterfall.

Until the shingles were ripped off, it was unknown what architect Brainerd Jones had engineered, or whether he had left that problem to contractors Williamson & McKenzie. Aside from the gutter profile shown above, there was nothing specified in the blueprints or in the written specifications. Thus we were completely surprised to find ourselves staring at not one, but two pipes emerging from each downspout. It was the oddest thing that any of us had ever seen.

Each pipe was 1½ inches and had bent-up lead flashing, which suggested that both pipes once emerged from the roof. The lower one was clearly once attached to the gutter; but what purpose served the upper pipe?

One theory was that the roof originally had oversized dutch gutters (also called box gutters, yankee gutters or integral gutters). These were sometimes used in Victorian-era America and were part of the house structure, built between the end of the roof and the flat board fascia that is visible from the ground (good diagram here, if trouble visualizing this). In this theory, our pipes acted as dual drop outlets (or maybe, a drop outlet with an emergency overflow drop outlet). Logical, yes, likely, no. The monster-sized fascia covering this design would have to be at least 8" high, and nothing like that is seen on the early house portraits or in the blueprints.

Theory #2: When the original gutter system failed, they improvised a second gutter above it to help intercept the cascade, and what we thought was a rainstop was actually a trough gutter (AKA stop gutter or Yankee Barn gutter). It's an interesting possibility, but couldn't be true; the two pipes were made and installed when the house was being constructed.

So to paraphrase Sherlock, once you've eliminated the impossible, whatever remains - however unusual - must be the truth, and this is it: Each downspout had its own private air vent.

LEFT: The northeast downspout and vent, as seen from the front
MIDDLE: The same downspout and vent, seen from behind
RIGHT: The center downspout, still sealed with a lead plug

Technically these are "back vents," and Brainerd Jones might have suggested that the contractor build these as part of his terse directions found on page 21 of the specs: "Vents from other fixtures, back vent where practical..." But the two big questions are A) was this design unique? and B) did it even work?

Researching this proved a great challenge; understanding turn-of-the-century plumbing is sometimes like trying to solve a crossword puzzle written in Ye Olde English. Terminology wasn't always the same as today (and sometimes descriptors meant the complete opposite), and there were a jumble of terms to describe the same thing. Quiz: Around 1900, what did they call the pipes that carried rain from the gutters? Possibilities include "downspouts," "conductors," and "leaders," sometimes mixed with a prefix of "roof" or "rain" or "water". When I discovered hyphens were sometimes used ("rain-water down-spout") and using that combination returned still more unique results in a Google book search, my brain popped a tiny aneurysm.

Victorians had the notion that a downspout could have multiple purposes. They could double as air vents for the sewer system, or rainwater could be diverted into regular waste pipes in the house, with the hope that a big storm would flush away grease and other clogs. Of course, if the downspout itself was clogged with leaves or snow, the entire venting system failed; sinks didn't drain and toilets didn't flush. That practice ended and you'll find many building codes from c. 1890 onwards adamant that "Rain water leaders must not be used as soil, waste, or vent-pipes." The emerging uniform building code also included two other points of interest to our mystery. There was a rule for downspouts concealed inside buildings, specifying that the downwpout be made of sturdy material and absolutely watertight with the gutter - a no-brainer, but it shows inner-wall downspouts were considered an acceptable building practice. And at least from 1915 on, the Uniform building code required vents for each plumbing fixture, but specifically stated no vents were necessary on downspouts. That might have been included for completeness' sake, or it might suggest that some builders were venting downspouts, and the practice had come to be considered superfluous.

But after all that research, I could find only a single mention of an actual vented downspout. It appeared in a 1889 journal published by MIT, then reprinted the same year in the widely-read trade magazine, "American Architect and Architecture" (Volume 26). It described an ancedote told by New York's Chief Inspector of Plumbing, where a basement was flooded because the downspouts couldn't keep up with high water volume from a heavy rain. The problem was solved by adding an air vent to the downspout. Now the idea makes sense; vents allowed water to shoot down those pipes like a firehose.

So finally we're at the big question: Did Brainerd Jones' system work?

Engineering reference books in that era provided tables showing every calculation a builder would probably need, including the proper diameter of gutter and downspouts calculated for the size of the roof. The rule at this time was one square inch of downspout for every 250 square feet of roof area. With each watershed of the Comstock House roof covering roughly 2,000 square feet, the house needed eight square inches of downspout per side. The vented system here came up short. Each 1½ inch diameter pipe provided about 1.75 square inches, so all three downspouts together offered a combined 5.3 inches.

The other part of the calculation was the size of the gutter itself, and here either/or the architect or builder stumbled. The charts specified a gutter of eight inches wide; the redwood gutters used on Comstock House were the standard five-inch width, which we verified after finding a discarded cut in a crawl space, seen at right.

But the undersized gutter wasn't the worst problem; we discovered that the middle downspout was never connected. The pipe was still sealed with a lead plug that would have been used during a plumber's smoke test for leaks. Considering the incredible speed of the construction of Comstock House, it's likely that the construction foreman didn't notice that the roofers had shingled over the plumber's mistake. The result was that the front downspouts could not have handled a watershed for a roof less than half this size.

Conclusion: The original rainwater system for Comstock House never worked. The gutters were too small, they needed an additional downspout, and one of the downspouts they had was permanently plugged. The rainstop was installed within a few years, but that didn't solve the problem either. Finally the entire system was abandoned; the redwood gutters were ripped out and replaced with a standard "ogee" five-inch sheet metal gutter, with three downspouts destroying Brainerd Jones' look. And despite all that, the gutters still couldn't handle the volume of water in even a moderately strong rain.

Hopefully our new system, which is over-engineered according to the period specs, will prove more robust. We'll see.

In the months leading up to the roofing project, most attention was over something no one will ever notice. Hopefully.

Besides the failing shingle roof that had us playing attic shuffleboard with buckets during every rain, there was clearly serious problems with the bargeboard - namely, that most of it was either gone or rotted.

Bargeboard is the trim along the edge of the roof; when it's horizontal, it's called a fascia board and is the plank of wood behind the rain gutter. But when it's on a diagonal eave or gable, the same piece is called a bargeboard, rake board or vergeboard, and its purpose is to cover structural details, like the exposed end of the rafter. In the Carpenter Gothic style, the bargeboards were often elaborate mouldings, such as lacy "gingerbread." Here are the before and after pictures of the bargeboard on the west face of Comstock House (CLICK to enlarge):

Our bargeboard is an odd, custom shape that none of our experts had seen before, and isn't found in the E. L. Roberts millwork catalog that was published in 1903, just before the house was constructed. The exterior of the board is rather plain, so it wasn't intended to be decorative; architect and preservation expert Mark Parry thought that Comstock House architect Brainerd Jones used the bargeboard to hide the edge of the shingles to prevent a ragged profile. But the side of the bargeboard facing the shingles has a deep notch with an unclear purpose, the mystery made even more difficult to solve because all of the surviving pieces were in very rough shape. On some boards, it looked like the notch was intended as a channel to carry away rain that might otherwise wick up under the shingles. Or was the original purpose of the notch to create an interlocking system between the bargeboard and shingles to seal the roof? Whatever the intent of the original system is now unknown; the wood is so badly eroded by more than a century of weather that precise interpretation is possible. Jones didn't write about it in his specifications to the contractor, and those who reroofed the house in later years made no effort to figure out the purpose, either roughly nailing the old bargeboard on the rafter ends as trim or replacing it with ordinary 1x6 planks.

It was contractor David Jessen who (as usual) figured it out. The flat surfaces on the back were meant to be attached to 30° furring strips, long lost. If the bargeboard is nailed to that and the whole assembly is attached to the end of the rafters, it creates a bed for the outside edge of the shingles. The roof gains a clean visual line as seen from the ground, and there is no exposed bottom to the shingle for possible rain wicking.

LEFT: The profile, with outline corrected for weathering
MIDDLE: David Jessen holds a test piece of the bargeboard with the furring strip
RIGHT: The barge and furring strip attached and sealed

To mill these custom pieces (and others), we provided moulding profiles to Redwood Lumber Company of Healdsburg, who can actually supply wood from old-growth redwood trees that were felled in the same era as the construction of Comstock House. (Unsolicited endorsement: Highest quality work at great prices.) The outline above left shows one of the best surviving profiles of the original bargeboard, with the believed original profile sketched in.

LEFT: David Jessen and Oscar Zavala test the gambrel roof angle
RIGHT: Except for the starter course of shingles at the bottom, the edges are completely hidden from the ground view

Newer Posts Older Posts Home