Introduction: Roberval Balance
What follows is first a set of notes, meant to be a complete explanation of how I made this balance. After that follow step by step instructions on how I made this balance. I do not expect the reader to be able to recreate exactly what I have done here. In that sense, this is not so much an "Instructible" as it is an "Historical". But it may still prove useful to anyone considering constructing such a device, so I have created it anyway.
The notes and the steps and instructions overlap in several places, and may even contradict each other— I want to get the instructions published more than I want to edit them to make sure they do not overlap. I hope that in the end they do not contradict each other too much, though they very well may. Anyhow, here are the notes:
What makes this kind of scale more interesting than a traditional beam balance is the ability to place the object to be weighed in a pan or dish without having suspensory chains or rods in the way and without regard to where on the plate the object is placed. Making it work, however, turns out to be rather more complicated than a beam balance.
In this design, I utilized several small pieces of prefabricated metal acquired on eBay as machine samples (the seller's name is moorish-idol). This was convenient mainly because I was able to be certain that anything originally manufactured by this company would be “exactly” what it appeared to be (holes would always be exactly straight and round; the sides of objects would always be exactly straight; the diameter of pins would be exactly the same, etc.). The only error, then, would be whatever I introduced through my own attempts to modify these parts. Very small variations in the process resulted in the inability of the device to function properly. I learned as I went. I do not think the eBay seller is still selling sample packs that include all of the necessary pieces to produce this exact balance, but I am not certain of this. The long steel arms with the balls at the end are among what I think are discontinued parts for their sample packs. These are difficult parts to produce by other means.
This Roberval balance consists of four beams each composed of two arms. Each arm is made of a rod, either of a type ending in a bulb which I pounded flat and drilled a 1mm hole into, or another rod of the same length with the bulb filed off to create a tip with a diameter smaller than the eyelet in my pivot pins. Each vertical column consists of a short vertical brass tube with a screw-on steel pointed tip with a 1mm diameter hole prefabricated in it and running directly through its center. I ran a brass pin through each of these holes to create my pivots. Then I inserted the steel arms into the ends of the pins: either I ran the steel arm through the hole of one side of the pin or ran the pin through the hole of the steel arm on other side (the diagrams will make this much clearer than my words). Each upper arm is joined to a companion arm at the center of the balance. These arms are either connected with a brass tube or with a pair of screw-together brass pieces. The lower arms are joined with a single pair of tubes through which I have drilled a 1mm hole and inserted a 1mm pin. Where the upper arms are connected with two brass mating pieces (see diagram). I have created a fulcrum from an X-acto knife blade which I have filed and cut to suit my purpose here.
The knife blade is supported by two razor blades which are themselves supported by two columns at a distance that allows the arms to be inserted between them. In the exact center of the balance is a pair of vertical brass bars which serve to hold the pin of the lower arms in position so that those arms cannot swing left or right as weights are placed in the pans.
The entire assembly is mounted on an aluminum bar that is 1/8" thick and 7" long. I drilled holes into this large enough to run bolts to hold the brass bars and the razor blade columns in position. I also drilled holes and created threads for four corner posts to hold the balance about 1.5" above whatever surface it was placed on (this was done mostly so that I could have space underneath the aluminum bar to insert bolts).
I have illustrated the sizes and dimensions of the metal pieces I used to create the balance. It is not necessary to use these exact pieces, of course— the things that are necessary are that the pivots be able to move absolutely freely, that the pins in them be exactly perpendicular to the arms holding them, that the vertical distance between pins be exactly the same on both the left and right sides of the beams, and that the pans be as perfectly level as they can possibly be.
For the balance's two pans I have used the stainless steel backs of two wrist watches. These were epoxied onto other metal pieces and attached to the tops of the vertical columns. It was important that the arms not be able to touch these pans when the balance moved— if the pans were not placed high enough, the arms would touch them when the balance reached the far lower extent of its swing. This could have been compensated for by decreasing the size of the pans, but as I wanted to maintain the “lip” of the backs of the wrist watches, resizing the pans was not practical. I was better off simply making sure the pans were mounted high enough that they would not touch the arms as they swung.
The precision of each element of the balance turned out to be rather critical in its final performance. Even slight variations in things like the angles of the pivot pins ended up making the balance “stick” at various places in its swing. Whenever this happened, I had to figure out what was causing the sticking and then figure out how to redo it so that it was more precise/ accurate/ smooth.
It is very important that the pivot pins line up precisely with each other in the same vertical column since most of the friction in the balance takes place around these pins. The more precisely they are aligned, the less friction they will experience as they rotate (as the pans go up and down). Even slight twists in the alignment of these pins will cause the pivot to seize up at some point in its swing and/ or will cause the vertical columns to refuse to be truly vertical.
In the end I spent the better part of three weeks creating this balance from scratch and making it work more or less correctly. I do not expect anyone to have the exact materials on hand that I did for making their own such balance— if you decide to try to make one, you will probably need to have your own source of materials or be able to manufacture them using resources that you have on hand. You may also be able to acquire the parts from the same manufacturer that I did, but the manufacturer did not sell me the specific parts I used: I purchased a number of sample bags at $1 each (including shipping) and was never certain exactly what I would get in each bag. This was kind of fun until I realized I didn’t have enough of one particular kind of piece, and then I had to order more bags and see what I got. I spent $15 or $20 on sample bags, and now have a large amount of unused pieces that will end up going to other projects. This was not an efficient way to get what I needed to make this balance, but it worked.
The two most-difficult parts of the construction of this balance were creating the vertical columns at the ends of the arms and creating the central connectors for those arms. I ended up deciding to create them such that the upper arms, when joined, would hold the X-acto knife blade in the exact center of the balance, and the lower ones would be inseparable by design. The balance can be taken apart by unscrewing the upper central brass pieces (only the piece with the “pointed tip” should be unscrewed, not the one with the eyelet in it as that one will not turn, by design).
Let me cover again the design of each vertical column. Each is made of a brass tube to which are attached two steel pivots. The holes of these brass tubes had to be tapped to accommodate the steel pivots, and the threads of the taps had to be an exact match and had to be exactly straight— if the threads were too tight, the pivots would not insert completely and then would not line up with those of the other column. If the tap was off by even 2 degrees, the pivots would not insert straight and then their pins would be off, etc. Nothing worked unless the holes were tapped perfectly cleanly and as perfectly straight as a human with the help of a jig can do! (the method by which I eventually achieved this was complicated, tedious, and I am sure unnecessary— I will not outline it here, but will give more info in that step later). All else led to failure (repeatedly).
Into each of the pivots is inserted a brass pin. Each pin has an eyelet at one end which holds the pointed tip of one arm; the other end of the pin is itself run through an eyelet of one of the other arms. These are set up to alternate top/ bottom and left/ right, so that each column has a pin with an eyelet on one side and a projecting end on the other (see diagram). When correctly assembled and joined at the center, these arms and pins should hold together without any additional support. In reality, I have attached small droplets of solder to the ends of the pivot pins to keep them from slipping off of the steel arms. I had to be careful that this solder did not run into the pivot— if it did, it would make the pivot turn less than freely (and therefore cause the balance to not function correctly).
Each of the steel arms is mounted in a brass fixture. The brass parts for the upper arms consisted of two parts, one male/ female and the other female/female, that will screw together. When screwed, these pieces create a single object with two threaded female ends. These openings are not large enough to accommodate the steel arms I was using, so I had to drill them just wide enough for the arms to fit. Even so, there was still some play in each arm’s setting. I stabilized these by adding solder to the hole and melting it with the steel rod in place. Although the solder did not bond to the steel, it did bond to the brass and held the steel in place quite firmly. I had to be extra certain that the arms did not “sag” downward in the melted solder, which as tricky— even a slight sag would be enough to cause the balance to stick. I had to re-set these arms several times. See the step where I discuss this for more information.
The bottom two sets of arms are each connected by a brass tube. I created this tube by sanding down a much larger brass piece that happened to have a hole in it that was the exact same size as the widest part of the steel rods (lucky me!). I was therefore able to put a steel rod in each end and be certain that they would be straight (this was much more difficult to do with the upper arms). I also added solder to these tubes to make them hold the arms securely in place.
In the course of setting all eight of the arms in their sockets, I had to make certain that the arms with the eyelets at the ends (4) were all of the same length, or rather that they were all the same distance from the center point of the beam. There were four arms to measure this way, and two of them were in brass settings that were prone to a certain amount of sagging (i.e., the upper arms). Getting these four arms (lower and upper, left side and right) to be the same length with their eyelets facing the same direction (actually, it is only important that the lower arms have their eyelets facing horizontally— the upper arms are more flexible), was not easy. I ran a 1mm drill bit through all four of them to hold them all in place, and heated up the bases of them with a torch to liquefy the solder. Then, as the solder softened, I tried to nudge each brass setting into place while keeping the steel arms aligned and preventing some of them from sagging. Technically, it is only necessary that the eyelets that are going to end up being located on the same side of the balance (left or right) be the same distance from the center point of the beam, but it was simpler to apply a uniform length standard to all four arms rather than try to do two-and-two.
Speaking of that 1mm drill bit. Drilling through stainless steel is not exactly simple, and my setup required me to create four arms with eyelets in the ends of them which I had to drill myself. I began by taking each rod and smashing it flat at the tip with a hammer. Once I was satisfied that I had it large and flat enough, I placed it in a torch until it glowed bright red. Then I cooled it. This softens the metal. Next I created a small dimple in the center of the flattened bit by taking a fairly large screw with a very sharp pointed tip, lining that tip up with what I considered to be the center of the flat bit, and tapping it hard with a hammer. This of course flattened the end of the screw, which could never be used twice for such a purpose, but it usually created a small dimple in the flattened end of the rod. I then reheated the now-dimpled end of the rod to glowing a second time, cooled it, and then drilled with a hand drill with my 1mm bit through the steel. The bit got dull, and required resharpening more than once, but it did eventually drill through the steel. Creating the dimple made a huge difference in my ability to create these eyelets. Without them, the 1mm drill bit slid out of position very easily.
It is important to note: the taller you make the overall balance, the smaller you must make the weighing pans, and/ or the longer you must make the arms, and/ or the higher you must set the pans above the vertical columns, and/ or the greater the distance you must have between vertical pivot points on the vertical columns. This is because a taller balance means the arms will have more vertical swing, and if the swing is excessive, it will cause the plates to come into contact with the arms, which is undesirable. A lower balance will not allow the arms to swing as far, and is therefore less likely to cause the plates to hit the arms. The exact amount of swing a balance will produce can be determined beforehand through careful diagramming (or through trial and error, which is the hard way but also works).
It doesn’t matter whether or not the balance “balances” left and right when you are making the arms. During the construction process, it never will. That is for later. What matters most is that the arms are correctly ALIGNED so that they can PIVOT freely at their ends points, that they are the same distance APART at both ends of the balance, that the vertical columns are indeed vertical at all positions of the arms, that the pins in the pivots are at right angles to the arms, that the vertical columns are the exact same height, and that the central upper pivot point be as FRICTION FREE as possible (i.e., the blade of a razor). Balance left/ right is for later.
I originally wanted some means to fine tune the balance, and considered placing a screw into one or both vertical columns on which I could attach a nut that I could screw in or out to make that arm heavier or lighter. However, the very principle of a Roberval balance is that no matter where weight is placed, as long as it is connected to the vertical column, it will not cause any change in the net weight of that side of the balance. This is unlike a beam balance, which responds nicely to such a weight placed at a greater distance from the fulcrum (because it is placed FROM the FULCRUM, not FROM a vertical column, which it does not have). A Roberval balance will require some other form of fine tuning its operation, other than the adjustment of a weight by a screw that comes out of one of the vertical columns. Instead of finding a way to do this, I simply added weight to whichever pan was lighter when both were empty until I got an even response from both pans. This does not allow me to fine tune the balance, unfortunately. It only allows me to balance it with certainty one time, the first time.
Other problems: I found that I could get a better “net” result by gently shaking the balance after I placed weights in the pans: finer differences in weight could be determined this way. However, doing this inevitably caused the fulcrum to shift along the razor blade edges on which it rested until the columns were no longer vertical. I have not found a solution to this problem. As carefully as I have constructed it, it lacks the sensitivity I desired for it to have, and lightly shaking it regains some sensitivity though this causes it to go awry for other reasons. Ideally, the fulcrum should not be able to slide anywhere at all; constraining it, however, seems to cause loss of sensitivity. It is at its most sensitive when it is at its most free to slide out of position, which it seems all too eager to do.
Part of the problem is that by placing my pans above the vertical columns, I have shifted the center of gravity of the whole device upward. This means that there is some undesirable tension being caused at various pivot points, however minor. But as one of my original design concepts included pans located above each vertical column, this was inevitable.
Step 1: Create the Vertical Columns
I made a jig out of some metal tubing that allowed me to tap both ends of the brass tube that makes up the middle of the short column. It was important that these holes be exact. I created the tap itself out of another threaded screw with the exact same pitch and thread size by cutting three grooves into it to allow the chips to fall away as it cut (3-in-1 oil for cutting lubricant). I made the tap as deep as the threads on the steel pivot pieces so that those pieces could mount as deep as possible and have their heads rest snugly against the brass tube. I then mounted the pivots into the tube, and ensured their openings were parallel by inserting a 1mm rod through each and seeing if they lined up. If they did, then the openings were aligned. If not, I turned one pivot or the other until they finally did.
Step 2: Modify the Steel Arms
My working pieces included eight steel rods that were 78mm long from end to end and 2.35mm in diameter at their widest. At one end of each was a small threaded portion which I removed; at the other was a 2mm ball which on four of them had to be converted into an eyelet. To do this, I pounded the ball with a hammer until it was a disk, heated the disk in a torch until red-hot to soften it, and created a dimple in the disk by pounding it once with the pointed end of a large screw. I then reheated it in a torch flame until red hot, and then drilled it with a 1mm hand drill bit. I used 3-in-1 oil as a lubricant for the drill. Stainless steel is very difficult to drill through, and I broke at least two tips while cutting these four eyelets.
The other four steel rods needed to have their ball ends ground off and converted into 1mm tips. I used a Dremel hand tool to accomplish this. The tips needed to be 1mm in diameter for at least the first 4 or 5 mm of their length.
Step 3: Modify the Brass Mating Pieces
Four of these steel rods (two of each type) needed to be mounted into pieces that could screw together and mate securely. I used a drill bit with a diameter very close to 2.35mm to drill into one of the female threaded ends of all four of the brass mating pieces I used for this purpose. If the piece had two female ends, only one of them was drilled; if it had one female end, that end was drilled.
*Amendment: I found that no matter how carefully I drilled these holes or how clever I was with mounting the rods into these pieces, I could never get them to be quite straight and so could never get the balance to swing without twisting one way or the other. I am afraid that the only way to get these pieces to create a perfectly straight bar is by using the same technique for the upper arms as I did for the lower ones: i.e., using the brass tubes whittled down from the pendulum-llike pieces, which are the only things that will guarantee me a perfectly straight fit.
Step 4: Modify the Brass “pendulums” Into Tubes
I had other brass pieces shaped like conical pendulums which I needed to convert into tubes to hold the lower four arms (two arms per piece). The steel arms fit very comfortably and firmly inside these pieces (the diameter of the opening was exactly 2.35mm). I used my hand drill and dremel tool to grind off the excess brass until I had a shape resembling a tube. I then drilled a 1mm hole through both tubes for a connecting pin to be inserted.
Step 5: Mount the Arms in the Brass Pieces
Each of the arms now had to be mounted into a corresponding brass piece, and had to be mounted straight in. Also, all of the ends with eyelets needed to be the exact same distance from the central pivot point/ fulcrum of the balance, and the ends with tips had to extend slightly further than the ends with the eyelets. I cleaned up the brass on the inside of the tubes created in the previous step using some steel wool wrapped around a small drill bit, placed some flux inside the tube along with some solder, and inserted the steel arms. The solder would not bond to the steel, of course, but it would bond to the brass and so function as a kind of cement between the two. I used a digitial calipers to ensure that I was getting the correct length on each arm.
I gave the mating brass pieces similar treatments with flux and solder, and inserted steel arms into them. In order to make sure all of the eyelet ends were the same distance from the fulcrum, I ran my 1mm drill bit through all four and then lined up the brass pieces at the other end until all were the same distance from the fulcrum. I then hit all of these pieces with a torch until the solder melted.
Here’s one trick I discovered for making the upper steel rods with the pointed ends mount very straight in their settings: although I had drilled one of the female ends of each of the four brass mating pieces to make it large enough to accommodate the diameter of a steel rod, the drill inevitable made the holes slightly too large— this meant that getting the rods to insert and be extremely well aligned with the axis of the mating piece and with the axis of the other arm to which it would eventually be attached was near impossible. So I after placing some solder in these drilled openings along with some flux, heating them with a torch, and getting them coated with a thin layer of the solder, I inserted the steel rod, which would usually still fit, but would be out of alignment. I then attached the mating piece to a second spare piece in order to give it some length, and mounted that piece in my power drill. Next, I turned on my torch and placed the brass piece with the solder in the flame. With my left hand I operated the drill, turning it slowly in a clockwise direction, and with the right I loosely held the end of the steel rod. I could feel with the right hand how off the steel rod was from center. As the solder melted, however, I could feel the rod loosen up and stop turning with the drill (the solder was providing a kind of hot lubrication). I then moved the piece out of the flame, still turning it clockwise slowly and holding the rod in what felt like the central position with my right hand. The moment the solder solidified, the rod would catch and begin turning with the drill again. And it would also be mounted almost perfectly straight. If I didn’t succeed the first time, I just placed the brass piece in the flame again and repeated the process until I had the result I wanted. The steel tip was so far from the torch that it never got hot. This proved very helpful for obtaining a well-aligned rod on at least that half of the overall beam. The other side of the beam, with the arm that has the eyelet, was simply a lot of trial and error with the torch and the solder. The eyelets themselves had to be extremely well aligned with each other, so the technique above could not be applied to ensure their linearity as inevitably the position of the eyelet would move up or down as the solder hardened and then would no longer be in the same position as the other three rods with eyelets. Of course, for the bottom two rods with eyelets, mounting them straight was simple because the opening in the tube into which they were inserted was exactly the same diameter as the rods themselves. The only two truly tricky rods to mount were those in the upper two brass mating pieces that had eyelets, as these had to be very well aligned and also the exact same distance from the fulcrum as those on the bottom were from the lower central pivot point.
Step 6: Create the Fulcrum
The fulcrum used in this balance is a #2 X-acto knife blade cut according to the diagram provided. It is installed with the sharp edges of the knife facing downward, between the brass mating pieces of the upper arms, with the screw of one set of those pieces running through the manufacturer’s oval of the blade and the other through the U-shaped hole that I cut to accommodate it (cut with a diamond grit Dremel tool wheel— X-acto knife steel is very hard, and cutting it this way was the only way to get through it). It is important that the new edges given to the knife blade be actually sharp and not flat— if they are only squared off, the balance will tend to be unresponsive to minor changes in weight.
Step 7: Create the “tuning Fork” for the Lower Pivot
It is important that the pin placed between the lower brass pieces be prevented from swinging left or right as the balance shifts position. This is accomplished here with a pair of brass bars with a thickness of 1.5mm, a width of 3.12mm, and a length of about 20mm mounted into the head of a large flat-head bolt which is itself mounted into a nut that attaches the bolt to the stand of the balance. The completed component resembles a tuning fork in appearance. The slot in the head of the bolt was almost exactly 1.5mm wide, and with a minor amount of filing of the ends of the bars I was able to get them to insert into it very firmly. The same effect, however, could be achieved a variety of ways— the way I chose allows the pin to hang freely below the fulcrum, experiencing no more friction than absolutely necessary for the balance to operate. The thickness of the brass bars was constrained by the proximity of the brass tube of the lower arms to each other: if the arms were too close (i.e., if the pin between them was too short) then the balance would not be able to pivot here and would stick. By keeping the brass tubes at least 2.5 mm apart, I could prevent them from touching the tuning fork as it held the pin.
Step 8: Create the Pans
Each pan is made from the back of a stainless steel wrist watch. To this I have epoxied two sets of other metal pieces, one of which has a hole in it that readily accommodates the tip of the pivot pieces. I had to be certain as I was letting the epoxy cure that the pieces were lined up in a direct vertical line— and variation from that would end up making the pan tilt once it was set on the pivot.
In retrospect, that might not have been the best idea. If I had allowed the pieces to set at a slightly irregular angle, then once I placed the on the tip of the pivot, if the pivot were for some reason not perfectly vertical itself for some reason, I could rotate the pan just enough to compensate for this. Instead, with a more or less perfectly vertical pan assembly and a slightly off kilter pivot, turning the pan did nothing to compensate for things. If it were also slightly off kilter, then turning it would have at some point made it perfectly level.
I also have considered ways to attach a small camera bubble level to the pan so that I can tell whether or not it is actually level on the pivot. I think that a short length of X-acto knife that was attached to as to extended out the back or side of the pan with the small level glued to it should do the trick. In the end, ultimately, as much emphasis as I have placed on the orientation of the arms and the pins and the pivots, it is the pans which must be level for the balance to operate correctly (and the pivots must swing freely— that is also very, very important).
It was difficult creating a pan assembly that would actually fit on top of the point of each pivot The sides of each pivot tip are slightly curved, so I had to come up with a piece which would still fit over these tips and, when fully seated, would not wobble or allow the pans to tip. The pieces I ended up using were the only ones in my collection of parts which had a hole I the correct place and of the correct depth for this purpose.
It was also important that whatever I used to attach the pan to the pivot, the attachment had to be broad at the place where it met the pan. If it was narrow here, then the pan would not hold to it securely and would invariably come detached. The funnel-shaped piece I ended up using was broad enough at one end to hold on to the pan securely and narrow enough at the other to hold the piece with the hole it in which would insert onto the tip of the pivot.
Step 9: Create the Razor Blade Fulcrum Supports
The balance “balances” on the edges of two razor blades. These must be mounted at a height and width which allows the arms and central mating brass elements to be inserted between the and be able to swing freely. The X-acto knife which has been modified for the fulcrum itself should rest one of each of its two ends on these razor blades. Razors provide the least amount of contact surface area for friction to interfere with the performance of the balance. The problem with using straight razors, as I have done here, is that they allow the fulcrum blade to move left and right— not easily, as a sharp blade placed at a 90 degree angle to another sharp blade will not usually tend to slip, but enough to be a nuisance and to interfere with the easy use of the balance. A more ideal solution would include some sort of V shaped groove that was shallow enough to allow the fulcrum to pivot freely to the left and the right but deep enough to keep the fulcrum from moving left or right at the same time. A straight edged razor does not do these things. If the newly created blade edges of the fulcrum are at the same height (and they should be) then the height of the razors must also be equal to each other and must hold the balance far enough off the base for the pin between the bottom arms which has been inserted into the tuning fork to not touch the slotted bolt at the base of the fork, as well as low enough to prevent the arms from swinging so freely up and down that they come into contact with the edges of the pans (smaller pans would allow for greater swing, but the size of the swing is not the measure of a useful Roberval balance). It may be useful to determine in advance how much swing the arms were going to have and make sure the pans would not contact the arms in the fully down position on either side. I am not convinced that razor blades were the correct or best solution to the problem of how to make the fulcrum turn. The original Roberval balances turned on a pin inserted through a hole, and I only avoided this because I feared such an arrangement would create too much friction. On the other hand, using razor blades was also fraught with its own difficulties as outlined above. Take your pick.
Step 10: Assemble the Balance
Assembling the balance was difficult, and I did it several times. Getting the arms all inserted into the correct eyelets and attaching the fulcrum through the top set and securing that fulcrum into position by turning just one of the two brass mating pieces just enough to hold it in the correct position without allowing either pivot to come out of position was tricky. I have no advice on this. Once the mating pieces were connected, the assembly was ready to be placed on the razor blades.
Inevitably, the balance would tip to one side or the other, even with the pans empty (I would not attach the pans to the pivot tips until after I was certain I could get the balance to operate correctly). This was due in part to my improvisation on one of the arms which made it heavier than the other seven. Also, the weights of the watch backs was not equal (having come from different watch manufacturers) and this also contributed to making one side heavier than the other. I eventually added small metal pieces to whichever side was lighter until I got the arms to just barely respond to what I had done. I then adhered these pieces to the underside of that pan (which had to be done judiciously, since the weight of the glue itself would cause the pans to again become imbalanced). I at one point considered trying to attach a screw with a nut to the lighter side and turning the nut to extend the weight of that side, but then a Roberval balance is designed specifically to keep that from mattering, and I gave the idea up. Fine tuning a Roberval balance is not something which can be done by shifting a weight attached to either vertical column. It can only be done by increasing (or decreasing) the weight of one side or the other, not by shifting existing weight to the left or the right.
Step 11: Construct the Razor Blade Pedestals
Each razor blade is held in place by a stand or pedestal. I inserted the blunt edge of the blade into the slot of a bolt, the same kind of bolt as I used to hold the brass bars of the tuning fork, and then epoxied this into place. I rotated the bolt into a split aluminum tube with a flared cone at one end (flared end down). The split in the tube meant that it was flexible enough to accommodate the width of the bolt securely without having to be tapped first. The other end of the tube narrows on the inside slightly right at the point where it also flares out on the outside. This opening was accommodating to the same kind of bolt, which I then inserted from underneath.
I had drilled three holes in the center of the aluminum stand, one dead center for the tuning fork and one on either side of it for the pedestals that would hold the fulcrum. I inserted the flared end of the razor blade pedestal into one of these and bolted it in from the underside. The hole for the flare was not an exact match— it was not conical, though it was too narrow for the cone to pass completely through it. this was ideal, because it allowed me to “tip” the pedestals either closer together or further apart and still hold them securely in place. I just unscrewed the bottom bolt slightly, adjusted the pedestal, and then retightened it. It tended to hold the new position well.
I made two of these pedestals and attached both from the underside with matching bolts. These had to be either the exact same height (unless one side of the fulcrum was cut higher than the other— in that case, flipping the fulcrum around could largely compensate for slight differences in the heights of the pedestals) and had to be parallel to each other from an overhead perspective (that is, the planes of the blades did not have to be parallel when viewed from the side, but they needed to be parallel when viewed from overhead). If not parallel, they might interfere with the ability of the central brass pieces to rotate freely; if not the same height, they would cause the arms of the balance to operate at an angle and therefore to bind or stick when swung.
Because the top edges of the razors were flat, the fulcrum tended to wander. Holding it in position would require that these have divots for the fulcrum to sit in, and the presence of divots would potentially interfere with the ability of the fulcrum to tip fully left or right. This is one major disadvantage of this type of arrangement over having used a pivot pin. A pin inserted into a hole rather than an X-acto knife resting on two razor blades would have ensured that the arms always moved in the exact same vertical plane.