First, all of the materials are gathered; in this case, ebony, bone, and old air-dried maple. The overall plan of the rose is laid out on paper, including the diameters of the completed center portions (ebony/bone only and ebony/bone plus maple "arrowheads").  A template of thin brass is also fashioned at this time for the maple "squiggles." 

For the arms of the star, a wide laminate of ebony and bone is glued up, slightly thicker than required for the width of each arm and slightly longer than the desired finished points. Individual pieces are sawn from the laminate and planed on the "show" side to the requisite thickness (approx. 1.6 mm). The non-show side is left rough for expediency and the fact that it will not be a gluing surface.

The maple pieces are sawn to the same thickness as the ebony/bone pieces and planed on the show side only (again for expediency's sake and the fact that the back will not be a gluing surface). Thereafter, they are cut into a triangle slightly larger than the template. (The pieces were taped together first to allow me to cut several pieces at a time.) 

Additionally, a "riser" for building the dome into the rose is fashioned from a penny and a nickel, which are stacked and stuck together and to the board with double-stick tape -- admittedly probably not an 18th century practice!

Next, each ebony/bone piece is trimmed to an angle just shy of 30⁰(1) so that they can be joined at the center. The edges of each angle are planed(2) at a slight angle(3) so that the whole piece will be "domed" when glued, with a rise of about 3.5 mm (or in this case, the height of the penny and a nickel stacked and secured to the board with double-sided tape). All the pieces are test fitted- to make sure all of the interior points align and that the corners fit together correctly.

The arms are still too wide and long at this point in order to have some wiggle room for material removal. The desired final length of each interior point is marked on the back side of each piece.

Shown here are the pieces dry-fitted. The drawn circle shows the intended dimensions of the final rose.

Note 1: Although in theory the angle should be exactly 30⁰ (360⁰ divided by 12), there is a very good reason to make it just a fraction less, which will be addressed in Step 3, below. 

Note 2: All of the planing is done by utiltizing the block plane upside down in the bench vice to act as a sort of "planing table." This allows for very controlled work and visibility, though it can be a little hard on the fingers at times. Suffice to say, the plane blade must be very sharp!

Note 3: The slight angles is in relation to the mating surfaces of the central angles so that they will join correctly in a domed position when the pieces are glued. 

Once the central angles are complete, the maximum length of each arm is marked out on the back of each piece and trimmed to just a little over that length. Recalling that the desired length of each interior angle is also marked on the inside, each arm is now carefully planed, with each side reduced to both marks and drawing down to a point. The whole is then tested for fit. (In the image at left, the arm closest to the penny -- 5 o'clock position -- was accidentally made too short and had to be replaced, as the ruler slipped when I was marking the overall desired length.)

Gluing was done in the following order, resting the glued pieces on the coin "riser" to maintain the correct angles:

1. Each arm was glued to another for a total of 6 pairs. The fit of each pair was checked before gluing and corrected with a file, as needed, to make sure there was a consistent fit when resting against the riser.

2. Two sets pairs were then glued for two separate sets of two pairs -- each comprising 2/3rds of a half circle. Ditto the above comments about fit.

3. The two remaining pairs were glued, each to a previously glued set to make two half-circles, ditto the comments about fit.

4. Once fully dried, each half circle was filed flat along the joining edge. As mentioned in Note 1 of Step 2, above, the interior joining angle employed was slightly less than 30⁰. I wanted to make sure that if there was any discrepancy, it would be possible to flatten the joining surface without digging into the length of the arms. By doing so, the center of each half was slightly pronounced; or said another way, the total angle was slightly less than 180⁰, allowing the removal of material from the center of the section to be jointed, without paring away any of the length of the arms.

5. Once each section was flattened along the joining edge and the fit was determined to be satisfactory when each half was rested on the riser, the two halves were glued together. 

6. Finally, the outer surface of each arm was refined with file and scraper so that there is a continuous surface between the pieces.

Now that the star is complete, a round piece of laid paper is cut with the help of a circle cutting tool to the diameter of the centrally joined portion of the star (in this case, 17 mm). This is then glued to the back of the star. Invisible from the front, this paper acts as insurance against accidentally separating any of the already glued arms and provides greater rigidity to the existing structure.  

(I did actually accidentally bump the pieces and break them apart before the paper was glued in place -- which was easy enough to reglue. The idea of adding a preliminary round of paper did not occur to me until after this accident. I do not know if the originals had this as well, and I suppose one might do without it just fine, but I can assure you that it did make the structure much more solid and provided for a great deal of peace of mind!) 

Here the completed and stabilized star is placed behind the sound hole to make sure the fit is correct. The ends of each arm are just slightly longer than required for the full diameter of the hole, which is intentional. 

The completed star on its own would make a suitable rose, and in fact one surviving Polish guittar by Szymon Gutowski, Krakow, 1811 has just that! In that instrument, the points of the star are let into a ring of wood that lines the circumference of the sound hole. (Thanks to Wojciech Gurgul for information on that instrument.)

Using the previously cut maple triangles, the brass template is traced on each blank with a freshly sharpened pencil to have a record of the desired final shape.

The shape was then roughed in first by cutting away a triangular area at the bottom of each piece. Next, the majority of the shape was created by using a round file, which allowed all of the bottom curves and most of the top curves to be completed. This was then followed by small flat and half-round files to finish off the shapes. The desired shape was considered complete when all of the pencil marks had been removed through filing, and each piece could be judged against the brass template to make sure it was correct.

NOTE: Each piece was left slightly larger than needed to fit the star. The final fitting is to be completed after Step 7.

As can be seen from an examination of surviving roses of this style, the wooden "squiggles" must have been dyed (and potentially finished) before being fitted to tbe star, as no evidence of dye is present on the bone pieces. In this case, I decided both to dye and apply finish to the maple pieces before gluing, as I did not want to use a finish on the bone and ebony (though one certainly could, I suppose).

Historically, the builders probably used whatever dye and finish was going to be used for the rest of the instrument, so they would match. In this case, I needed to match the existing finish, so I built the color up using Trans Tint NGR water/alcoholo based dyes in multiple applications of a variety of colors, including "honey amber," a custom mixed orange, red, and "brown mahogany." Some traces of this can be seen on the cardboard work surface, shown behind the finished pieces.

Once each piece dried for over 24 hours, I applied multiple coats of finish.

Once the finish on the maple pieces was thoroughly dried, I began to fit them to the star. You will notice a piece of blue painter's tape on one of the arms of the star in some of the images. This was my "six o'clock" position. I wanted to be sure that I knew which maple piece went between which arms of the star, so I roughly sketched some lines on the cardboard and placed the pieces as they were finished in these areas to help me keep track (which can be seen in this photo).

Despite trying to be as accurate as possible with the angles, it was simply not possible for all of the spaces to be identical. Therefore, I started with the widest gaps and used the widest maple pieces first. Each piece was carefully planed on the two edges to be glued (which helpfully removed any finish and provided a clean gluing surface), correcting the angle as needed to get a proper fit at both the "arrowhead" and at each "leg." (The plane was used the same way as in Note 2 of Step 2, above.)

Each maple piece was then quickly glued in it's proper space, starting with every other one. Glue was applied quickly to the outer edges of the "arrowhead" and "legs," briefly slide into place then out again, glue reapplied, and piece slid home. Any excess glue was quickly removed with a warm, damp rag, and the end of the "legs" and the points of the star were held in place with plastic spring clamps until set (though not dry). In this manner, I was able to glue one maple piece about every hour and had completed half of the pieces after 6 hours.

One may notice from the picture that the "legs" of the maple pieces project far beyond the points of the star. This excess length was useful for handling the pieces while dyeing, finishing, and gluing them, but it will not be a part of the final rose. If one were to draw the angles out, it would soon become apparent that it is not possible to fit the next set of maple pieces without overlap and interference. Therefore, I trimmed the exterior of each "leg" where it projects beyond the point of the star, which can be seen in the next photo. (It is probably most apparent in the wooden piece in the 11 o'clock position.)

In theory, each wooden piece had been specially fitted to its exact spot and should be ready to glue in. In practice, I discovered that the end of the star's arms are somewhat flexible. Each remaining unglued piece had to be re-checked and in some cases replaned to assure a proper fit.  In fitting the remaining pieces, I felt it was important not to undermine the integrity of the rose by making the pieces fit too tightly or too loosely. I fit each one to be snug enough to stay in place when pushed "home," but loose enough that it did not distort the shape of the rose or try to push itself back out when fitted.

The remaining pieces were glued in place one at a time at the same pace of one an hour, in order to let the glue set. All told, it took a little more than 12 hours to finally have all the pieces glued (though most of that was wait time).

At this point, the whole piece is somewhat delicate. The joint between the sides of the "arrowheads" and the junction of the star arms is probably the strongest area, but still subject to fracture. Accordingly, I measured out the final diameter of the circle that contained both the bone/ebony parts and the maple parts (in this case, 27.5 mm), and using the circle cutter like in Step 4, a larger circle from is cut from laid paper.

Once glued in place a dried, I wrote my initials and the year so that future conservators/repairers would know that this was not an original piece of the guittar.

A quick word about the strength of this arrangement: As I mentioned, the whole rose was quite delicate when first assembled, but after the paper was glued in place, it demonstrated an incredibly rigidity. So much so, that despite the fact that some of the "leg" / "arm" joints breaking loose, there was no sense whatsoever that the rose would come apart. This, paired with the domed structure, created an incredible strong part that I couldn't help but poke vigorously just to test its strength!

Now that the whole rose structure was assembled, it positioned in the sound hole and temporarily taped in place with blue painter's tape. (At this point the 6-o'clock position was no longer marked by blue tape, since my initials on the paper were written in such a way that they would be right side up when the rose was in the correct position.) A sharp blade was then used to mark the wood just inside the edge of the sound hole.

All excess wood from the "legs" was then carefully removed using first a chisel, followed by a file, up to the marked line. The fit was tested multiple times and refined with the file by holding the fitted piece up to the light and looking for gaps or areas to had not been taken down enough. The goal was to fit the end of each set of "legs" to the inside edge of the sound hole. (It should be noted that since the instrument is 250-ish years old, the sound hole is not perfectly circular -- if it ever was...)

I got so excited about progress at this point, I forgot to take a photo!

Since the rose will be held in place by small wooden tabs from inside the instrument, and since the doming of the rose makes each piece be slightly angled, it seemed to make sense to flatten the back edges of the rose where they would be glued. This was one of my few nods to the modern age, as I accomplished this by placing a pieces of 180 grit sandpaper on the table of my bandsaw and gently rubbing the rose back and forth while turning it. The end result was a space of about 3-5 mm of flat area where where the backs of the "legs" will come in contact with the tabs.

Before any next steps, the old remnants of tabs were removed and the area cleaned with warm water and a toothbrush and dried. It was fun to see that indeed the old tabs lined up with the "legs" of the new rose, giving further credence to the idea that this was the type of rose the instrument previously had. 

Before gluing in the new tabs, which were made from offcut pieces of soundboard wood, around 2.5 mm thick, the new rose was positioned in its proper place, and pencil marks were made on the soundboard as to the size and position of the new tabs. Some of the tabs needed to be a little narrower because of the final size of the legs.

The new tabs were then cut and quickly glued in place with about a 3mm protrusion into the sound hole. Excess glue was carefully removed and cleaned from the front of the tab (which will eventually take the rose "legs"). Once the tabs were fully dry overnight, the rose fit was tested again.  Any excess wood that could be visible after the rose would be glued in place was carefully marked, the rose was removed, and excess wood was removed with a chisel. Thereafter, the front and visible sides of each tab were dyed very dark brown so that the light wood would not be strikingly noticeable once the rose was glued in place.

Rather than trying to glue all of the "legs" to tabs at once, I first glued the 6- and 12-o'clock positions, which just so happened to have the best / tightest dry fit. Clamping was not necessary for this, given how good the fit was. Next, I glued each subsequent arm based on its fit, starting with the ones that fit best first. Despite the overall rigidity of the rose with the paper in place, there was still a little bit of flexibility in the "legs," so some of them had to be "coaxed" a little more than others. Each subsequently glued "leg" was done in this manner, until all of the legs were glued:

1. A thin pallet knife was kept in the water bath of the hide glue to help it retain heat.

2. A small bit of glue was picked up by the pallet knife then deposited from the back between the rose "leg" and the tab.

3. The "leg" was held in place while a short loop (~40 mm) of narrow twine was draped over the "leg" from the front so that two loops could be seen from the back.

4. [Photo] A small hooked weight (200 g) was hooked through the loops, and the instrument was then placed on risers on the table so that the weight could hang free. The weight essentially pulls the twine against the "leg" so that the "leg" bears on the tab. (The instrument was propped at the heel and end block to allow room for the weight to hang.)

The fully glued rose is quite strong. The domed construction resists downward pressure (accidental or otherwise!), and any pressure on the rose makes it bear both against the tabs (to which it is glued) and the edges of the sound hole (to which it is not).