Friday, December 21, 2018

Testing casein paint on leather, part III

Dammit.

Our test board cured for an entire month.  Last week I added some thin layers of carnauba wax paste, which is, admittedly, not something that would've been available at the time (carnauba palms are native to Brazil), but I was attracted to the fact that it's harder than beeswax and has a higher melting point.  It's difficult to see, but on some of the swatches it dried to a whitish waxy appearance.  At this point, my favorite was the glue-soaked, sanded leather with a wax topcoat and no gesso (lower right corner), which was smooth and a bit glossy.

I wanted to first test the paint against rough surfaces and sharp implements to see if it would flake.  It didn't, though the knife and half brick scuffed the paint applied over unglued leather badly.  The glued/sanded/waxed/un-gessoed swatch looked to be the most durable, as both implements glided over it leaving superficial scratches.

The real test, of course, would be the rain.  I waited until today when it was convenient to set the board outside for an hour, as I figure these are realistic conditions.

In the meantime, I cut up a whole bunch of undyed, semi-bleached linen swatches, as I figure these would be the most likely to show paint smudges if someone's clothing were to come into contact with wet paint.  When the board came inside, I rubbed one swatch of linen against each swatch of paint.

At this point, it became difficult to read the labels, but that didn't matter.  NONE of the paint swatches adhered well enough for practical use; each one smudged badly onto the linen.

Sadly, the glue-soaked but un-gessoed swatches fared by far the worst, as the glue absorbed water and turned liquid.  The paint slid right off of them.  The gessoed ones didn't do much better.  The wax topcoats gave no noticeable protection.  In fact, the leather swatch with no treatment at all (corresponding to the linen at the bottom of the leftmost column) did best - or, at least, least worst - although as you can see, it still smudged.

So where do we go from here?
The thing is that milk paint is normally applied to raw wood, and it's said to be durable in this application, particularly if coated with a drying oil.  This leaves us with several options:
1)  Bare wood scabbards with only a drying oil or wax finish.
2)  Painted wood, again probably having a water-resistant finish.
3)  Unpainted leather facings, perhaps vat-dyed in a single color (I am told that a finish of beeswax and drying oil will prevent dye bleeding, but that's a question for another day).
4)  Embossed sheet metal facings, either over a leather facing or glued directly to the wood.
5)  Oil paint or encaustic (wax).  Unfortunately, I know of no evidence that these types were used in our period.

The first two leave us with the issue of the scabbard possibly splitting along the seams when the blade is pushed in or jostles inside the scabbard while the wearer is running.  Only the chape would provide reinforcement.  It's possible that small bands of organic material such as rawhide were tied or stitched around the scabbard farther up, leaving most of the wood bare for painting, but that's pure speculation.

Options 3 and 4 have good historical grounding, as the Issyk Golden Man's scabbard was covered largely with leather and small sheet gold adornments, while the Achaemenid scabbard from Egypt had a (now-vanished) gold facing pasted to the wood.  We're still left between a rock and a hard place:  Most reenactors won't have the skills, or money to pay someone who does have the skills, to make a full metal facing, since fitting one to the complex curves of a wooden core in the classic Achaemenid shape would require lots of careful hammering to expand the metal in certain areas.  On the other hand, leather alone is pretty dull-looking, and I've never seen evidence for Achaemenid scabbards with partial metal plates such as those at Issyk which would be easier to make.

I'll look into carrying out milk paint tests directly on wood.  In the meantime, Ivor at RAT brought to my attention the book Chariots in Ancient Egypt, which is said to have some evidence for painted leather.  Now I have a few more weeks of free time before the spring semester starts, I can finally start checking whether it contains anything potentially applicable to our situation.

Friday, November 16, 2018

Testing casein paint on leather, part II

Traditional gesso is sold as a powder, which is mixed with water in a 4:3 ratio, covered and allowed to soak for a day before warming it to melt the glue.  Warming produces a texture like melted ice cream.  The liquid is then painted on, allowed to dry, sanded, and painted on again.


After the first coat, I found that the remaining gesso cooled to a bubbly, rubbery texture that could be ground into the preceding layers, squishing it into a smooth paste in the process.  Both wet and dried gesso is a stark, eggshell white.


The next step was to mix and apply the powdered milk paint.  This is a sample of Real Milk Paint Co.'s "barn red," which has a slighly dull, ochre-like color (though if I use this type of paint on a large scale, I plan to use a colorless base and add storebought pigments so I know they're authentic).  I mixed it up in a common salad dressing-style cup with a lid and used another cup to measure out the water, as the recommended ratio of powder to water is simply 1:1.

Not long after the paint was applied, most of the gessoed swatches showed fine cracking.  Presumably the gesso absorbed water from the paint and swelled.  The swatches with glue but no gesso didn't seem to have this problem.

Interestingly, the swatches that were neither soaked in glue nor gessoed seemed to do very well, with no cracking even when I tried poking at them with my fingernail.  This may change over the following month if the curing paint becomes harder and more brittle.  The other factors are how well it handles water damage and whether the paint over chamois can be sufficiently sealed with wax if water becomes a problem:  a hard, non-sticky wax may also tend to flake off if the underlying leather is still puffy.

Saturday, October 13, 2018

Background for testing casein paint on leather

Casein paint, or milk paint, is made by mixing milk protein (usually concentrated or coagulated) with pigment and often hydrated lime, which causes a chemical reaction that makes the dried paint harder and more durable.  Roman reenactors frequently use it to paint their scuta (large legionary shields) and report that it's fairly water-resistant, more so if a wax top coat is applied.  The basic ingredients should have been available in Achaemenid Persia, and the use of milk paint is documented from the Bronze Age, so I consider it a reasonable option for period-appropriate paint.

Could it have been used on tanned leather?  I'm planning to make some more akinakes scabbards - well, sort of; I'll have to farm the wooden cores out to a professional woodcarver this time because I no longer have access to the kind of tools that are needed to produce them in a reasonable timeframe.  Anyway, wooden scabbards are greatly strengthened with a facing or other reinforcements to keep them from splitting along the seams, but aside from the chape (when it was made as a separate piece), the only reinforcements known on akinakes scabbards are facings and the only facings known are leather and embossed sheet metal.  Shaping sheet metal around the expanded throat of an Achaemenid scabbard requires great skill, so I'm going to rule it out.  That leaves leather.  As I've said before, I believe the usual leathers used in the period were fat-cured (similar to braintan and chamois) or rawhide.

Chamois is cured with fish oil, just like buff leather and German buckskin.  In the U.S. it's made from sheepskin, although the term chamois in French refers to a species of goat.  As a result, chamois is thin, elastic and soft, so it should be easy to stretch over a scabbard.  Its non-abrasive properties make it popular for polishing cars, so it can be bought in automotive and hardware stores.  I bought a three-square-foot piece from Ace Hardware for $20; I think this should be sufficient for covering two scabbards.

Milk paint is generally said to need a ground (the surface onto which it's painted) that is both hard and porous.  It works very well on raw wood.  On vegetable-tanned leather it's sometimes reported to crack.  However, as far as I know, nobody's tried it on suede leather before, so I wonder if it might not be able to soak in like thin hide glue paint.

An alternate method might be to harden the leather by soaking it in hide glue.  However, I believe this would reduce its porosity.  There are two methods of prepping a non-porous surface for being painted with milk paint that I read of:  sanding the surface to roughen it and give the paint something to "bite" into, and priming it with gesso, which is a mix of glue and chalk dust.

RAT member Feinman informs me that soaking the leather with hide glue would make it hygroscopic (meaning it would tend to absorb atmospheric humidity), which could present problems for paint adhesion.  I know of no way of dealing with this.

The other factor is whether or not to apply wax.  Over bare suede, I think that wax would tend not to seal very well.

For the time being, I intend to test everything.  I've gotten a wooden plank and attached small swatches of chamois with hide glue, each of which will then have some milk paint applied.  There are nine swatches, the three largest of which will be shared by two test treatments, so as to try every combination of leather glued only by the back, soaked with glue, sanded or not, primed with gesso or not, and waxed or not.

(The nailed-down rawhide pieces in the middle were from another project.)  As you can see, the glue-soaked chamois is brown and almost transparent.  It dries hard enough that I can barely dent it with my thumbnail, although it's not as hard as rawhide.  Also, chamois that's glued on the back but not soaked and then allowed to dry can be soaked later by rubbing more glue on the front.  The glue seems to bond the chamois pretty strongly to the wood.

Over the next few days I'll have to narrow down the possible choices of gesso and wax.  Once all the paints are thoroughly cured (allow a month) and treated, I'll test the paint's adhesion when dry, adhesion when wet, and resistance to bleeding and smearing when wet.  Stay tuned!

Sunday, September 30, 2018

Plataea 2021

Two days ago, Christian Cameron of Hoplologia and the Amphictyonic League announced a tentative date of June 28 to July 4 to mark the 2,500th anniversary of the Battle of Plataia, one of the largest battles and major turning points of the Greco-Persian Wars.  Get your arrows ready and join us!

Wednesday, May 23, 2018

Cast belt concho

Another item I cast this spring was a concho for a weapon belt.  I've done a number of posts on this subject and I won't go into the background yet again, so here goes.

As with most of my bronzes, I modeled it in Victory Brown.  I formed the cone by rolling out a sheet of wax to roughly 1/10 to 1/12 inch/2-2.5mm thick, cutting a circle, removing a small pie wedge from it, and bringing the cut edges together.  I cut the scalloped edge with an X-Acto and formed the rims around the petals by rolling ultra-thin wax "noodles" and pressing them into place, separating the petals by dragging the tip of a hobby awl between them.

On the back, I attached six wax gates of around 1/8 inch/3mm near the edge and two of perhaps 1/16 inch/1.6mm a bit in from the edge to serve as the rivets (as no rivets are visible on the original's face, I assume they were cast in place on the back, since adding them and then grinding them smooth on the front might undo the peening).  These all branched out from a central main gate of about 1/3 inch/8-9mm.  There were no gates or vents on the model's face.

The back bar was cast straight as a section of one of the thicker gates, with room on either end so angle grinder could cut through without touching either the bar or the concho itself.  Since I had doubts about whether it would cast, I didn't poke holes all the way through for the rivets before casting, but instead just made small pits to serve as guides for the drill press (which has an annoying tendency to slip a bit to one side when first entering).  I then drilled fully through the pits after de-gating, and formed the kink through repeated bending with needlenose pliers and annealing.

BCCC lab tech Nick investment-cast the model using a centrifugal casting machine.  This technique, also called spincasting, allows for very fine and detailed models which are almost guaranteed to cast whole, unlike gravity casting, as the centrifugal force throws the molten metal into every nook and cranny.  It also obviates the need for venting, since the weight of the molten metal being sent into the mold forces the air out the same way, through the pouring gates.

One drawback is that the investment, which as I understand it is basically plaster of paris with some sort of strengthening agent, traps a lot more small bubbles than ceramic slurry does.  Where these bubbles are in contact with the wax, they form little warts all over the model, making finishing more tedious.  You can see one large bubble on the bottom center there, and close examination will reveal the remnants of a number of smaller ones.

On the plus side, the investment is much easier to remove than ceramic shell.  Most of it crumbles away with a dunk in water, that which is left can be softened if necessary with a baking soda solution, mostly scrubbed away, and any remaining traces scraped off with metal tools or buffed off with a wire Dremel brush.

The rims around the petals are almost certainly too prominent.  Going by the OIP drawing, I interpreted them as sort of a gently rounded lip, but because I didn't spend enough time smoothing the noodles down, they instead look high and sharply differentiated from the petals' faces, with crevices that I couldn't successfully polish.  As a result, there's still a significant amount of brown as-cast finish there, giving the face a slightly antiqued appearance.  The dip around the middle cone is probably too deep as well.

Dry-fitting the back bar was the most tedious part of the process, since the rivets were crooked where the met the concho's back, and I still failed to get it perfect.  Once the fit was sufficient, peening the bar into place only took a couple minutes.  I did so with the face resting on a thick carpet on top of concrete, so as not to damage it.

As you can see, after being bent, the back bar sticks out considerably more than the original.  I should have made it only a little bit longer than its desired final length.

As I mentioned before, I think this is a low-strain application, and the thin bar and small rivets will be good enough.  Should they break, a replacement can be silver-soldered on.

If possible, I intend to make a whole new belt with a strap of buff cowhide in place of my old belt's veg-tan.  I will probably still use latigo lace for the time being, however, since it doesn't stretch like fat-cured leather, which is likely to become an issue when supporting the fully-loaded gorytos.

Wednesday, May 16, 2018

Cast arm fibula

Three years ago, I fabricated an arm fibula out of brass rod.  I was never satisfied with the lack of depth afforded by filing the details into stock that was narrow to begin with, so I've always wanted to replace it with one that was properly cast.

Last semester I took foundry at BCCC again and took the opportunity to make a few more reenactment items.  This one, which I finished up just last week, is wax-cast (as the originals probably were) and based on an example found at the Harvard Art Museum.

Although this is a huge step up from the brass one, I still made a couple of mistakes and compromises.   First, I had been meaning to replicate one from Deve Hüyük, and ran a simple Google search for this purpose, but didn't read the article carefully enough.  This example is of unknown provenance, although the article does point out that it's similar to two Deve Hüyük examples.   All three are of unusually large size, each around 4 inches/10cm long, with "bead and reel" decoration, flared at the base, and having fingers filed in after forging.

I cast it as a blind vent on a larger assembly.  The hand was cast as a bulb.  Unfortunately, I didn't account for shrinkage, resulting in a mushy hole where the wrist should've been, and the bulb fell off during de-gating.  I forged the bulb out into a flared catchplate roughly 1mm thick (finding out, along the way, that the Everdur silicon bronze Bucks uses actually hot-forges nicely).  Luckily, crack lab technician Ray was able to weld the catchplate back on, but the result is a much thinner "wrist" than the original has (it's still thick enough, though).

You'll also note that the proportions are slightly off.  This probably had to do with my inability to sculpt the wax for the beads and reels thin enough, and of course I bent the elbow a little too sharply.  The rough patch on the elbow is a remnant of the chill ball, and I may grind it smoother later on.  The original might have been cast straight and then bent, but I don't think this is the case, since it's slightly thicker where it's bent than immediately above or below.

The other problem that occurred with casting was that the slurry, which is quite thick, apparently failed to flow between most of the very thin wax noodles which formed the reels on my wax positive.  The first pair of reels just below the elbow cast fine. The second cast part of the way.  The rest cast as single thick reels with only a short groove in one or two places around.  This was sufficient as a starting place for me to file the grooves in, but it means that the grooves are sharp and the sides a bit square, unlike the nicely rounded cross section on the original.

The fingers were filed before bending the catchplate over.  It proved too difficult to get the bend in the catchplate to align with the pin, but as you can see, this was a problem with the original as well, so it's perfectly accurate.

I made the hole for the pin with a drill press, unfortunately a bit off-center.  The pin itself is simply a piece of springy 3/32-inch steel welding rod hard-soldered into place.  The original used bronze, but the bronze welding rod at Bucks was too thin and I imagined that brass would contrast unpleasantly with the orangeish Everdur.  In any case iron was also commonly used for arm fibula pins.

Monday, January 8, 2018

Conversion reversal

More than two years ago, I undertook a project of modifying half a dozen Native Way CH257 tanged triangular Chinese arrowheads into socketed Scythian-style trilobals.  The result was acceptable in weight though a bit crude in appearance.  However, when I shot them at Marathon 2015, the heads had a tendency to lodge in the wooden target while the rest of the arrow popped loose, as the arrow glue by which they were attached failed.  Arrowheads with the tangs left intact did not come loose.

I was left with two options:  order more arrowheads and convert them into trilobals without removing the tangs (arrowheads of this type are known from Achaemenid Persia, though rarer than the socketed types), or try to attach new inserts with a stronger bonding material.  In this case I chose to create new tangs by soldering the socketed heads to metal inserts.

Since the sockets were created with a 1/8-inch drill bit, it was easy to create the tangs from 1/8-inch brass rod.  I sanded the ends of the rod segments up to past where they would enter the sockets to make sure they were clean and free of oxidation, smeared a little solder paste (top) onto the tangs and placed the arrowhead onto the tang.

I worried that the tangs would anneal and wind up bending if they struck a hard target.  To try and prevent this, I clamped each tang up to 1/4-inch from the point in a heavy vise before hitting it with the gas torch, and applied heat only just until the solder melted and flowed around the junction.

The soldering seems to have been successful, but whether it will be strong enough to withstand shooting into hard targets remains to be seen.  With a melting point of 430F, this is not a "hard" solder, and the contact area can't be very large.

I also took the opportunity to anneal and straighten the tang of a Native Way G202.  While I have not yet tested it, I think that crooked tangs may run the risk of bending further when striking a hard target.  On the other hand, tangs that have been softened too much through annealing may also wind up bending.  These questions will have to wait until spring for answers.