The Griffith School of Music was located in Atlanta, Georgia. It was founded and directed by Mary Burke Griffith (photo below) who ran the school with her children out of their three story home.
Her children all married and eventually all lived and worked together. The three children who worked in the school were: William Butt Griffith (1880-1964), his wife, Margie Keelin Griffith (1891-1965), L’Ella Griffith Bedard (1883-1971), and Mary Griffith Dobbs (1890-1970).
She also had a child that did not work at the school. His name was Beverly Griffith and had several interesting jobs to include, hotel management, movie appearances, stunt driving for films, directing films, war correspondent and director of public affairs for Eastern Airlines.
The school sold, taught and repaired all stringed instruments, and were representatives and dealers for the Gibson Mandolin and Guitar Company. The Griffith School was on Peachtree Street before it moved to 650 Bonaventure Avenue, prior to 1928. The family lived up-stairs and the teaching studios were down-stairs. There was another house behind the school, which was William Butt Griffith’s studio.
William Butt Griffith was the original owner of the mandolin now known as “The Griffith F5”. He was the director of the Atlanta Mandolin Orchestra, President of the Atlanta chapter of the American Federation of Musicians, and was well known for being an instructor, performer, conductor and instrument dealer.
Photo below shows Margie Keelin Griffith, owner of the unique Gibson A5 mandolin, seated behind the harp. Standing behind Margie K. Griffith is L’Ella Griffith Bedard. Standing beside Mrs. Bedard is William Butt Griffith, holding his famous F-5 Gibson mandolin. Alas, he is standing behind Margie’s harp.
Margie Keelin Griffith was also the first harpist of the Atlanta Symphony Orchestra and organist at the First Church of Christ Scientist for over 40 years.
Photo below shows Mary Griffith Dobbs all the way to the left, and Margie Keelin Griffith all the way to the right.
Margie Keelin Griffith would teach mandolin from time to time, and would borrow her husband’s Gibson Loar signed F5. As the story goes, she loved the sound of his mandolin, but the points on the F5 mandolin bothered her leg, so she asked if they could order one without the points. The mandolin was completed and the label signed by Lloyd Loar in 1923. Serial # 74003
William Butt Griffith passed away in 1964 and Margie Keelin Griffith passed in 1966. They had no children together. Tut Taylor bought both the F5 and the A5 from Mary Griffith Dobbs sometime after Margie passed but before the school closed in 1966. After the school was closed, Mary Griffith Dobbs and L’Ella Griffith Bedard sold the house and moved elsewhere.
Photo shows Mary Griffith Dobbs all the way to the left standing:
The A5 was played by Norman Blake on John Hartford’s Steam Powered Aereo-Plain which was released in 1971. It can also be heard on the Steam Powered Aereo-Takes album released in 2002. Also check out Tut’s album “Dobrolic Plectral Society” which the mandolin is featured on track three, “Griffith Mandolin Society”. I’ve only been able to find this album on vinyl.
At some point in the 70’s Tut sold it. Around 27 years later he was able to “have it back for a while”. In 1994, David Grisman and Tony rice released Tone Poems with the Griffith A5 on track 9. The mandolin was put up for sale in Oct 2010 for the asking price of $345,000, but as far as we know, it did not sell and stayed with the current owner who is located in Washington state.
We were contacted by Samuel Hardman who was the last student to receive a teacher certificate at the school.
He studied there until 1966 when the school closed.
I was the last student of the Griffith School of Music to receive a Teacher Certificate. Mine was in piano, theory, and harmony.
I also received a certificate in music history. Further, I was taught to play the harp and pipe organ. I was able to do so because I was a good music reader and player before I attended the school. My first day at school was a splendid day in March. As I walked up Bonaventure Avenue, spring flowers were in bloom and a light wind was blowing from the south. I could hear the beautiful sounds of harps and I knew that I had found my heaven! Every minute at the school was a pure pleasure and I fully intended to please my superb teachers.
William Butt Griffith was the first member of the Griffith family to die when I was in school. He was my harp teacher. After I complete my lesson, we stood talking for a while. It was late in the day when I returned home. About an hour after I left the school Mr. Griffith became very ill. A Christian Science Reader was called. Alas, he soon made the final change with all the members of the family present. Mr. and Mrs. W. B. Griffith had no children. After William Butt Griffith departed, his wife seemed to become very weak. Indeed, soon she did not teach and went away for some months to rest. One day I went to school and after completing a lesson, I was told that Mrs. W. B. Griffith was in bed and that she would not answer when called. Members of the family wished me to go up stair to her apartment and to check on her condition. I did so and found her in a very deep sleep. John O. Mitchell was called and Mrs. Griffith was removed to Piedmont Hospital. She made the final change late that day.
The school was full of Gibson instruments that were no longer used. Some months after Mrs. Griffith departed, Mary Butt Griffith Dobbs, sister of William Butt Griffith, placed some of the instruments on exhibition at a music store in Decatur, Georgia. Among these was Mr. William Butt Griffith’s F-5 mandolin. One morning before going to school, I received a call from Mary B. G. Dobbs. She wished me to pick up the F-5 from the music store and bring it to school. I did so. Later in the morning, two men came to the school. I was soon informed that they were there to purchase Mr. Griffith’s mandolin and went out into the reception room to be with Mary B. G. Dobbs. One of the men purchased Mr. Griffith’s mandolin. When they were told about Mrs. Griffith’s unique mandolin, they wish to see it. I was sent upstairs for the mandolin. I think the same man purchased it. L’Ella Griffith Bedard and I thought it very wrong for Mary B. G. Dobbs to sell the instruments, but we could say nothing about the matter.
My final recital at the Griffith School of Music was in April 1966; however, I continued my lessons with L’Ella Griffith Bedard until the school closed. Indeed, I was there the last day and locked the doors myself. It was the end of something very precious.
Emails from Samuel J Hardman
OK, so now we have some details into the history of this mandolin. I am friends with a couple people who have had access to the instrument, so as new info comes in, I’ll be sure to update the history page.
The next thing we are going to do is go through drawing up the plans for a “copy” of the instrument. I chose to start with the peghead overlay. Reason being, my friend Bruce Harvey from Orcas Island Tonewoods was able to obtain a direct tracing of the peghead outline from Bill Halsey, a bow maker out of Michigan. Here is a photo of the template that Bruce mailed to me:
The first thing I need to do is import this picture into my drafting program (I use Rhino 3D) and trace it.
An interesting point on the overlay is that it is not symmetrical and we will be replicating that in the final product. Here is a shot of the peghead folded in half so you can see the difference. Subtle, but definitely there.
Here is a picture of the overlay traced and cleaned up in Rhino………
Bruce had this really good idea, to inlay “The Griffith” into the peghead of these copies. I had him send me a picture of the inlay that he had made for his project. Again, I traced that in Rhino then sized and placed it on the overlay.
Here is a whole peghead shot with tuners drawn in. We will be using Gotoh tuners as they are the closest looking to the original. They are not perfect, but close. I also drew in the classic Fleur-de-lis
Here is a picture of the original peghead…………………….
For the fret board, I choose to take the measurements directly from Adrian Minarovic’s detailed Loar signed Gibson plans. A few interesting points about the original fret boards… Scale length on most modern mandolins is usually set to 13 7/8 ths. The original Loar fret scale was advertised as 13 15/16th, so this is the scale length we will be using on this project. This fret board also had the small fret wire common at the time which measured around 0.034 wide by 0.032 high. This size fret wire is no longer in production so we decided to go with the smallest available Jescar fret wire which measures 0.040 by 0.039. The binding on the fret board measures 0.040 wide, and that is a common size available today. The frets terminate at the edge of the ebony and do not overhang the binding, another detail in which we will replicate. The dots will be mother of pearl sized at 0.25 Finally, the fret board will be milled flat true to the original. On future copies, I will install radiused boards with larger fret wire if the customer requests it, but for the first two, we are trying to get it as close as possible to the original. Here is a picture of the fret board drawn up in CAD.
For the next step I need to draw up the plans for the rim. My friend Harvey Marcotte happens to own a Gibson A2Z from the same year that this instrument was made. Since the Griffith rim would have been made from the same set of forms, I feel this is the most accurate way to get the correct body size.
So to do this, all I used was a couple of blocks and some card stock with a hole cut into it to get an accurate tracing. Since the back is arched, it is impossible to lay it on a table and trace it. Here is what I came up with.
After the tracing was made, I drew in a perfect 1 inch square then brought it down to my local office store so that I could scan it, them import it into my drafting program. Here are the results.
The scan was imported into Rhino (autocad) and traced for the final result:
There is no way to get accurate drawings of the blocks used unless I had access to the actual instrument and a CAT scanner, or if I had access to a disassembled snake head which I do not. I decided to draw in the actual size tail block used on an F5 of the same time period. The head block was an average of two sets of plans and visual inspection of the block on the snake head. I have drawings, from published plans, of head blocks that were similar but each were somewhat different, and neither of them matched the outline of the actual snake head, so I took some artistic liberties and drew it in the best I could from all the data I could collect both in print and visually. I don’t suspect any minor differences in the neck block shape will alter the tone of the final product.
Update to previous post: After posting a link to this blog on facebook, I was contacted by Darryl Wolfe who offered to send me some pictures, info and a partially disassembled A2 from the same year as the Griffith build. Darryl is the author of the F5 Journal and has an immense amount of knowledge on the Loar signed Gibsons. I’m very grateful for the additional information he provided, thanks Darryl!!!
With that said, I was finally able to get an accurate tracing of the blocks that would have been used for the Griffith mandolin. Here is the partially disassembled A2 that Darryl sent as well as the blocks that I was able to trace and duplicate. I had initially thought that the tail block would be similar in size to the tail blocks used for the F5’s of that period, but I found that the tail block on the A body instruments was slightly larger.
Here is the inside of the A2:
Here is the head block I made. Note, I am unable to lay it flat due to the locating pin on the A2 headblock, which is why that gap is visible.
Last, here is the tail block:
Next I’ll get started on the bending form that I’ll use for bending the sides to make the rim. Here is the drawing for that. I’ll cut this out of plywood or MDF.
It is believed that Gibson used steam and forms to bend and clamp the sides of the instruments. I don’t have a steam generator so I’ll be approaching this the same way as we bend our sides. A form and an electric heating blanket. I could bend it over the hot pipe, but I find the heat blanket and form prevents kinking and cupping in the wood which saves tons of sanding and unnecessary sanding of the wood thinner than initially planned.
11 June 2019
Today I finished building the rim bending form. I decided to try something different than what I drew up in the last post. The idea behind this new design, is that it will use quick clamps and overall be lighter than the previous design and different from what we use on our modern A5 for bending. This is my rim bending station. I use a controller which heats up the blanket and maintains it at a specified temperature.
But first, I have to select some appropriate side stock. The Griffith has relatively plain sides that are flatsawn. For Bruce’s instrument, I’m trying to get as close to the original look as possible. On the second instrument, I’m using wood that has more figure. This is the board I selected for Bruce’s stock.
Notice how the Griffith sides are similarly plain.
Next step is to thickness sand on my drum sander.
One of the sides will then be put in the form and lightly clamped. Once the temperature of the blanket reaches 200 (detected by the thermocouple wire) I’ll start the bend.
Next, I take a block and push, then clamp the head block side of the rim into place, followed by the tailblock end. I then heat the rim slat up to 300 degrees and keep it there for 15 minutes. Once my timer expires, I turn off power to the blanket and let the wood cool naturally to room temperature.
Once cool, I take the rim out of the form and repeat for the second rim slat. You can see that the rim slat holds its shape nicely and fits perfectly on the head block with just hand pressure.
Once the rims are cut to size, we can glue the blocks in. I’ve used both inside and outside forms and the inside forms and the inside form seems to do the best job.
When the hide glue has dried, I then move on to installing the triangular kerfed linings. On our modern A5’s we build, we use either reverse kerfed lining or heat bend solid linings. I prefer either of those over the triangular kerfed linings for several reasons. With the reverse kerf or solid linings the glue up can be done with regular pony clamps. The completed rim is also more rigid and holds its shape very well, so I don’t have to keep it in the form until the top is glued on. However, since we are replicating what was used on the Griffith, we are using triangular basswood linings for these builds.
This is our kerfed lining glue up station:
Kerfed linings being glued up using some new clamps I found at Stewmac.
After the linings are installed, I wait a day then level the linings to the rim slats and the rim is now complete. It will stay in this form until the top is ready to be glued.
Switching gears, I felt like working on fret boards. A cool story about where the ebony came from:
One day I got a call from my friend Rolfe Gerhardt from Phoenix mandolins. He said, “Hey, I might have found some ebony so meet me down in Massachusetts and we’ll have a look”. We drove to the address provided and pulled into an old wooded lot with a house and a couple sheds on the property. Rolfe introduced me to the individual who was the son of an engineer that used to work at the old Ovation factory in Connecticut. Back in the day, if ebony was not completely pitch black, it was discarded. This engineer collected all the discarded fret board blanks from Ovation and kept them at his house for years. He ended up passing and his son took the whole stash. His son wasn’t a builder and didn’t know what to do with it and ended up finding Rolfe’s info. There was probably about 1500 sq feet of storage space, all filled up with ebony fret board blanks, guitar woods, mahogany blocking material and other odds and ends. We purchased a large number of the blanks, but I’ve been told that a large guitar manufacturer has since bought the rest of it. We will be using blanks from that stash. While not jet black, we have dyes that we can use to make it that way.
I thickness sand the blanks down to a few thousandths proud of 3/16th. I then mill the fret slots, position marker holes and then cut out the profile all on my CNC machine. Gibson used to use gang saws that would cut many fret boards all at once in a fraction of the time it takes my machine to cut them. An interesting note about some of the old fret boards from the 20’s is that the fret slots are in the wrong position!! Many of the Loar signed instruments have had the fret boards replaced due to this. When the gang saws went dull they took the blades of as well the spacers between them. The size of the spacer is what designated the fret slot distance. Somehow they got mixed up and reassembled wrong. The resulting fret boards had some of the frets in the wrong location. That won’t be a problem with our boards, accurate to a thousandth of an inch! Seasonal movement of the wood itself introduces a greater deviation.
The boards are now ready for installing the mother of pearl position markers and getting bound with 0.040 ivoroid binding with a side stripe.
For the back, we decided to make a plaster cast of the A2 that Darryl let us borrow. Once cured we can use our machine and probe it to make a digital copy of the plate. I did some on line research and found a page that detailed how a violin builder copied plates of old violins, so we followed this making a few changes in the size of the materials to accommodate a mandolin. That page can be found here.
First we cut a piece of plywood and drill a hole in it to allow the air to escape when we pour in our plaster.
The back plate is then placed on the backer and a foam dam is cut out to surround the plate. A 0.006 thick piece of latex is placed on the plate prior to clamping the foam dam.
Plaster is then mixed and poured into the dam. A plywood cookie with holes is installed to help stabilize the cast.
After 30 minutes, the whole thing is flipped over and the plate removed. Ready for 3D probing on the machine after a couple days of curing.
Here, we have removed the plaster cast from the foam dam and placed it on the machine for probing.
Here is a quick video of the probe touching the cast, it was set up for collecting points every 0.040 of an inch.
The machine took 26 hours to probe the whole plate. Here is the resulting point cloud as shown in my CAD program.
A surface is made using the point cloud and it ends up looking like this. Lots of clean up still left to transform it into something we can use.
Now that I have the plate scanned in I need to trim the waste points and create a surface that the machine can read and duplicate. I create a surface in my CAD program and push and pull the surface until it matches perfectly. In this photo, you can see where the mesh (created by the points) and the surface are not matching. I spend untold hours messing with this to get it correct. Believe me, it is actually easier to just go ahead and do this by hand carving the plates with gouges, planes and scrapers. However, I want a repeatable carving that will match as close as I can get it to the orignal plate.
After many weeks of working with both the top plate and back plate scans, the models are almost complete. I drew in and attached the heel button on the back because the snakeheads didn’t have this feature, but the Griffith mandolin does.
Now we are ready for some test carving using MDF (medium density fiberboard). Everything looks pretty good. The edges have a flat designed into them that goes around the whole plate. This is so I have a nice flat surface to run against my router when I’m making the channel for the binding to sit in. Once the binding is installed, I will fair the recurve into the binding to make a nice smooth transition. More on that later.
For the top, I used the main arch profiles that I was able to generate from the A2 plate. I had emailed Darryl about a picture he had posted on the Mandolin Cafe a while back on how he thought Gibson might have approached this. It is believed that they would have used a plate that was originally carved for an oval hole and re-purposed it for the Griffith mandolin. This would make a lot more sense to do it this way than to completely redesign a one off instrument. It would also explain the oddities found in the Griffith top plate, like the bridge not being at the apex of the arch. Below is the picture that Darryl posted.
From looking at this photo, you can see how the area under the fret board was carved off. I was able to do the same thing for my top plate by modifying the original scan, but maintaining the rest of the arch profiles to remain as true to the original as I could. The A2 in the picture is the same A2 that Darryl loaned me to use for this project.
Now I must select the correct wood. Below is a copy of the actual spec sheets that were found in the Gibson office at some point and were posted over on the cafe.
There is no way to verify the actual species used on the Griffith, but through experience and conversation with wood dealers and historians, I’ll bet that they did use “Adirondack or W.Virginia spruce” which really means Picea rubens. There is also the posibility that it could be a Sitka or Engelmann based on the spec sheet below that calls out the wood quality of tops. The A models called for wood that was from “Oregon” which could be either Sitka (Picea sitchensis) or Engelmann (Picea engelmannii). But maybe they made the provision to use Picea rubens?
The top on the Griffith has that muddy appearance the stain makes when the wood is off quarter. How off quarter is hard to tell, but it could be anywhere from 20 to 45 degrees off quarter. An interesting note here is that this wood would be rejected by most modern builders. In fact, while obtaining the wood for this project several years back, I found it very hard to source off quarter wood. When I asked several dealers why they don’t have it, they responded that most builders would reject it, so they usually just burned it up! To show how off quarter some of these Loar signed instruments were, I’m posting below a photo of a CAT scan that was shared by Adrian Minarovic over on the cafe.
For the maple, I’ve worked with enough sugar maple to be fairly confident that the back on the Griffith is an off quarter sugar maple (Acer saccharum) back. Sugar usually has a very distinct look to it, and that piece just screams sugar. However, I can’t be 100% on this, but for our project we are going to stick with sugar maple, as that is what it looks like was used on the majority of the Loar signed instruments.
Once I’ve selected my wood, the first thing I need to do is pass it over the joiner. This removes the milling marks left by the saw mill.
Next, I’ll heat up some hide glue that I make from granules and water, heated to 140 degrees.
One of the halves is placed in my shop vice and I use a 5 1/4 hand plane to remove the scallop marks left by the joiner. No matter how well setup your joiner is, it will always leave scallops which are tiny areas of compressed wood. This won’t make for a suitable joint so it is cleaned up with the hand plane.
Once both haves are planed, I check the fit in the vice that I’ll use to glue up the plates. Looking at the glue build up on the vice jaws, you’ll see I use this setup often. Most of our mandolin plates are joined this way.
Next, I heat up the mating surfaces with a heat gun, apply a bead of glue and then use what a technique called the “rubbed joint”. Because the spruce comes in wedge form, it is difficult to clamp. By rubbing the plates together, it squeezes out the glue, and then you will feel the glue “grab”. That is when you know to let go and walk away. Hopefully your plates grabbed in the right position so you maintain your book match. This is how violin builders have been gluing plates for the last 300 years. I use this method on the backs as well.
After about a half an hour, I’ll move the plate to the adjacent drying rack and repeat for as many plates as I need.
After 24 hours of glue cure, I’ll plane down the glued billets to the approximate thickness I’ll need before machining. I usually use a small 13 inch planer to accomplish this.
Next I need to make my fixtures for holding the plates while they are being machined. They are made from Medex, which is similar in apperance to MDF, but doesn’t contain formaldehyde and has much better stability. I machine these out on my CNC and then they are sprayed with a couple coats of home made shellac. A rubber gasket is installed into a channel I route into the perimeter of the plate and a hose barb gets inserted into a hole that is drilled so that I can pull a vacuum on the plate to hold it down.
Next I’m ready to machine. But before I get to that, I’ll make mention of the fact that this is really not too dissimilar to how it was done in the 20’s. What?!?! you say? Well, it’s know that Gibson used a pantograph router, which were popular back then for carving chair seats. A stylus would ride a “buck” which was the carved blank of what they wanted to produce. There would be a series of shaper heads that would ride on a rail, driven by belts, that would duplicate the movements of the stylus riding over the buck. Depending on how many shaper heads, they could replicate many parts in very little time. Basically it was a manual CNC, and many builders I know use the same idea, but on a much smaller scale than the old Gibson factory would have used. It is know known how many shaper heads they had on their equipment as most of that info has been lost, but I heard that 10 head units were common, meaning that they could produce around 10 plates at a time, which most likely took under an hour. On my small machine it takes about two days to run 6 plates.
I install my first vacuum fixture that I made on my spoilboard.
I then use an acrylic template I made to drill registration pin holes on my plates so that I can accurately mount it to the vacuum fixture ensuring that the center line is true.
The blank is then mounted onto the fixture and below my machine I fire up my vacuum pump.
The plate is sucked down to the fixture. I then install additional clamps to hold the whole thing down. I’ve had too many times where the vacuum broke loose due to the porosity of the wood, which the machine would then toss the free plate across the room, usually destroying the plate and anything that it hit!
Then I run the two programs that I wrote after designing the plates. One is a roughing pass using an end mill, and the second pass is made using a ball mill. Please note, that these videos are drastically speed up.
I then change vacuum fixtures and get ready for the machine to hog out the inside.
I have the machine carve it so that it is just under a quarter inch thick. This is how it comes off my machine, and I’m confident that this is similar to how a plate would have been given to a craftsman for final graduations at the factory.
Next I need to graduate the plates. Bruce shared the graduation map taken by John Sullivan many years ago.
No two pieces of wood are the same, and I don’t know if the stiffness or flexibility of the Griffith plate, or if the thickness on this map will work for my plate. I’ll make that decision based on my experience and how the wood is behaving. There is however an overall graduation theme here that I’ll follow.
First, I’ll sand the outside profile smooth up to 120 grit. As mentioned previous, I’m leaving that flat around the edge which will serve as a guide to route my binding channel. Once the binding is installed, I’ll fair the recurve to the outer edge for a smooth transition.
Next, I’ll establish the depth of the recurve on the inside of the plate with my 10mm finger plane.
I then establish the depth in the approximate center.
The two depths are then blended together with the finger planes.
A scraper is used to remove the large tooling marks.
Finally I sand the plates to get a smooth surface that is now ready for tone bars.