by Riley Doty, Doty Tile
A year later I’m picking up the topic of setting Subway Ceramics tiles with 1/32″ joints. I recently returned from the large national tile trade show, Coverings. There I met Jan Hohn from Minneapolis who won a major NCTA/Tile Letter award for her installation of period floor tiles set with 1/32″ joints. She used the Victorian/Edwardian type of tile called “Geometrics”, the undecorated single color unglazed tiles which are associated with decorative encaustic tiles of the period. The installation is in a sun room or breakfast room of a c. 1900 house and the layout of the colorful Geometrics make up a carpet like pattern which is true to the very best installations of its day. The juxtaposition of colors is accentuated by changes from straight grid to diagonal, using triangles, rectangles to other polygons, in a complex repeating rhythm. Jan is also experienced at setting Subway Ceramics wall tiles and I’m hoping she will write her opinions on this blog. I’m sure she’ll have both agreements and disagreements with my various statements. One of the methods she has used is an old one in the tile trade – setting wall tiles with string joints. Maybe she will describe how she uses that method!
Why 1/32″ joints are not approved by the tile industry:
Prior to a discussion of setting with these tiny joints the increased danger of cracked tiles must first be pointed out. I have installed a couple of large bathrooms with Subway Ceramics, using a maximum size of 1/32″ joints. I think the result is strikingly more authentic, more similar to period work than the same tile when set with a wider joint. But this is NOT a method that is approved by the NTCA (National Tile Contractors Association). Their reasoning is valid. With tiles butted so close together the installation has diminished ability to absorb shock or movement without cracking. (Even cementitious grout joints offer a bit of stress relief within a field of tilework.) In the next paragraph I’ll discuss how I view this practice which violates industry recommendations.
Examine some wainscot walls that are 100 years old. Few if any are pristine. There are generally two defects in evidence. (a) Cracks appear every so often, running across the field of tiles. (b) There are also usually cases of spalling – edge chipping of the glaze – on some individual tiles. These are the same defects most likely to appear in closely set fields of subway tiles today. (The same physical laws apply!) So in one sense if you want thoroughly authentic subway tile walls you may expect and accept some cracking and spalling. I personally aim to minimize their occurrence and would prefer to avoid cracking and spalling completely, but some appearance of these defects should realistically be considered acceptable. There are some measures which can be taken to try to minimize stress in these closely set walls, and discussion of that subject will have its own section below. (At the same time please note that using the recommended 1/16″ minimum grout joints will NOT ensure that the installation will be free of cracks! Tile is brittle and many forces have the potential to cause cracking – so this is a relative matter of following recommended practices in order to diminish the likelihood of cracking.)
My installation procedure:
As I’ve said earlier I prefer to use a medium bed mortar with a 3/8″ x 1/4″ square notched trowel, and I like to presoak the tiles. I am going to assume that the final height can “run wild”. By this I mean there is not any exact mark that the top tiles have to hit, rather the work can proceed upward until it finishes at whatever line the tiles need to maintain full units. Obviously the module is 3 1/32″, so there is an opportunity at approximately 3″ intervals to have the tilework end with a full uncut course at the top. (Occasionally the work must be planned to finish at an exact predetermined line, but let’s leave that possibility out of the equation for now.)
A. Overview:
(1) I stack off a level aluminum L-edge which I screw into the wall a my predetermined grout line, two courses above the bottom
(2) I set the wall in horizontal bands, working up from the bottom, each one being four courses in height
(3) When I have smeared and set a band of four courses high I stop to adjust and level it.
(4) I observe whether the highest point in the top course of tiles falls at or below the 12 1/8″ line which was my smear line.
(If the high point is above 12 1/8″ then I will have to re-calibrate and adjust by making 12 5/32″ my new layout module for each section.)
(5) After the four-tile-high band is set I put up a metal straightedge and rest it on the tiles.
(6) I secure the L-edge with blobs of thinset, press it in, and set a level on its flange.
(7) My level indicates which is the highest point.
(8) I put red wedges (the kind with razor thin tips) in until the top of all the tiles on the fourth course touch my L-edge..
(Some tiles will be slightly larger than others so some may not require a wedge, but will simply butt each other.)
(9) I remove the L-edge, cleaning off the thinset that held it to the wall.
(10) Before tiling further I first mark my future smear line for any adjacent walls (by laying an L-edge from the top of my finished section).
(11) I proceed up the wall this way in bands which are four courses high.
(12) I like to use a straight piece of lightweight wood about 3/4″ width to “true up” the tile faces, both rubbing with it and tapping on it with a rubber mallet.
(13) Periodically I hold up the 3″ edge of a tile to check my offset, to make sure the running bond doesn’t drift off center.
(14) Please see the earlier blog, including about presoaking and having to remove a portion of tiles to butter them out, etc.
B. Further detail:
-I usually start setting on the longest wall. First I will establish a level starting line. Usually this means that I screw an aluminum L-edge onto the wall. Whether or not the floor is level I want to start stacking off a level base. (I will later take away the L-edge and cut in to the bottom). I then mark and strike lines for three sections (at 12 1/8″, 24 1/4, 36 3/8″ ) each section being four courses high. I like to do sections of four courses rather than three because it is easier to manipulate the trowel and comb thinset for a nominal 12″ band than a 9″ width. These lines will probably be my permanent marks, but that is provisional until after I see that the tiles can actually fit within those tolerances. As I said in #3 above I may have to switch to a larger module if the tiles tell me so.
-I have an assortment of L-edges, with no more than 6″ intervals or gaps from one length to the next. They are generally full of holes drilled from past jobs but I am always ready to drill any new holes in new spots needed to attach them on the present job.
-There is a certain calculated risk that after one’s module dimension has worked for initial sections that a later batch of tiles may turn out to run a little larger in size. One reason I try to start with the longest wall is that this improves the odds that I will have a mix that will include larger tiles. However it may be wise to compromise slightly and work with a 12 5/32″ module right from the start; the visible different will be slight and this helps insure some wiggle room in case larger tiles do appear later on. It is true that Subway Ceramics wall tiles do vary more than 1/32″ from large to small. (The industry recommendation is to establish a grout size that is three times the maximum size variation of the tiles being installed. So we understand that what we are trying to do here is decidedly not pragmatic or necessarily easy!)
-DO NOT UNDER ANY CIRCUMSTANCES AGREE TO SET THESE TILES USING A 1/32″ FOR THE SAME PRICE AS NORMAL TILEWORK!!
-It is wise to work out of several tile cartons at once, shuffling them like a deck of cards (because the tiles in certain boxes may tend to run larger than average.)
-In case you do run into larger tiles on an adjacent wall which will not fit the module used on the long wall then you have a minor emergency on your hands. Sometimes things go smoothly for awhile and then one runs into a change in average tile size. Larger individual tiles will have to be removed from the field and swapped out for smaller ones. Not fun, but never as bad as it first seems when panic sets in – there will only be a finite number of oversized tiles and the problem will get resolved.
C. Possible means of accommodating stress:
-I am an advocate for installing “mini expansion joints”. This is not something which is recognized officially in the tile industry so I will describe it here: This is a scaled down version of standard expansion joints in larger areas of tilework. It is modeled after the use of perimeter expansion joints on floors which ARE strongly endorsed by tile industry bodies such as NTCA and TCNA. Under that system, foam strips either 1/4″ or 1/2″ thick are installed around the perimeter, wherever tilework butts up against hard restraining surfaces (such as walls or building columns). When the finished installation undergoes stresses of tension and compression these soft areas give the tiles room to safely expand. One beauty of perimeter joints is that they are often hidden by baseboards, etc. In those cases the installer does not even have create visible joints – which require considerable work and present a challenge aesthetically in terms of making them blend in nicely with the grout joints.
The material I use for “mini” expansion joints is strips cut from foam sheeting which is slightly thicker than 1/16″. I get the foam from Home Depot for example where it is sold in rolls for use under laminates that use “floating floor” systems. My reasoning is that given the size of most residential tile installations that these appx.1/11″ strips provide a meaningful expansion zone. (Further they ensure that there is no hard connection between a tile field and any adjacent restraining surface. Thus if there is structural stress perpendicular, or in any direction relative to the tiled surface, the tilework will remain unattached and unaffected.) This thin film serves some of the function of an old fashioned cleavage membrane or “slip sheet”.
-Before I install backerboard or mortar I put a strip of foam film on the face of the studs on the perpendicular side walls – afterwards this will be trimmed back flush to the edge of the board. Similarly when I turn to work on the perpendicular walls I put the foam on the surface of the backerboard or mortar. And later when I install the tiles I first apply foam film to the perpendicular restraining surfaces. When the work is finished the corner joints will receive matching caulk instead of grout. Foam is similarly installed at the bottom, between the floor and the wall tiles, and this joint is also caulked.
I have a definite feeling this system is not destined to catch on widely, but I suggest it because it is a possible way to mitigate some of the potential for tile damage from stress. I have no engineering knowledge to quantify how effective this maneuver may be, it’s an intuitive thing. I would be interested to see it tested but meantime I use it in many installations.
I also have a strong preference for using Schluter KerdiBoard as my backerboard of choice, and for working with 4′ x 8′ sheets to minimize joints in the substrate. In closing I want to emphasize that I am extremely satisfied with the results of two large installations I did with 1/32″ joints in 2006 and 2008 respectively. Both are still free of cracks. I hope they will stay that way. If they crack to a slight extent during my lifetime I think I would still favor the tight joints personally, but if they were to crack extensively I would probably abandon the practice of setting these tiles so tightly. The most important thing though is to preview this with the clients and make sure they understand to risk involved.
{ 0 comments }
