Introduction Turning slender wood spindles can be challenging. Generally, for any given spindle profile, as one tries to make the spindle slimmer and slimmer, eventually the spindle begins to vibrate and whip under the pressure of the cutting tool, leaving a rough chatter-like finish on the wood. The onset of this effect can be delayed, but not prevented, by steadying the spindle with one hand and balancing out the tool pressure. It is well known from experience that some mounting methods perform better than others in terms of their resistance to this effect. However, no one seems to have tried to quantify these differences. In an attempt to fill this gap, I recently undertook an engineering analysis of the subject, augmented with experiments to validate the theoretical calculations. As a practical matter, mathematical analysis is not feasible for arbitrary spindle profiles, so I have chosen to restrict attention to simple cylindrical spindles. While the results obtained will not be exactly correct for any particular profile, I believe that they will reflect the relative performances of the various mounting methods on the average over many different profiles. The results obtained suggest that the choice of mounting method can have a large impact on the ability to turn long slender spindles. Mounting Methods The following mounting methods were considered, each classified according to the standard beam design problem that it represents: 1. One end supported in a chuck, the other end free. (Cantilever beam.) 2. Supported between centers. (Simply supported beam.) 3. One end in chuck, the other supported by a live center. (Beam fixed at one end, simply supported at the other end.) 4. Both ends supported in chucks or equivalents. (Beam fixed at both ends.) 5. Same as (4) but under tension. (Pre-stressed beam fixed at both ends.) I did not pursue the first method in detail, because in practice where there is likely to be a problem, the turner usually starts off with method 3 and parts off at the very end of the process. Other possibilities that I did not cover include the use of a steady rest with any of these mounting methods. Performance Characteristics The rigidity of these mounting methods, that is, their resistance to deflection under tool pressure, varies widely. Also, in each case there is a particular point along the spindle where the rigidity is lowest. All other factors being equal, finish turning of the spindle should begin at this location and work outward toward the supports. Table 1 is a summary of the calculated performance characteristics of the first four methods. Calculation of performance for Method 5 is not practical and was left for experimental determination. I considered three measures of merit. The first is the rigidity of the set up, that is, the tool force needed to produce a given worst-case deflection of the spindle. The second is the minimum diameter of spindle that can be achieved for a given length of spindle. The third is the maximum length of spindle that can be handled for a given diameter. I have assigned to Method 2 an arbitrary value of 1.0 for each of these figures of merit, because it is the most common mounting method for spindle work and will be used as a standard for comparison for the other methods. The rigidity of each mounting method was calculated from standard beam deflection equations. The rigidity of a cylindrical spindle is proportional to the fourth-power of its diameter and inversely proportional to the third-power of its length. Therefore, having calculated the relative rigidities of the various mounting methods, I was able to calculate the relative minimum diameters and maximum lengths that just allow a given amount of deflection. The third and fourth columns show the results of those calculations. The fifth column shows the points along the spindle, measured from the driven end, at which the rigidity is least. Finally, the last column shows the axial tension on the spindle. Best performance requires that any compression on the spindle be very small. This is of particular concern in Method 2, because unless an aggressive drive center is used, considerable compression force will be needed to provide the needed drive friction. I ran a series of experiments to try to validate my premise that rigidity is a good predictor of performance in spindle turning. In each case, a wood dowel of fixed length was mounted according to one of the mounting methods and turned down in successive passes over the whole length until chatter began to develop. Care was taken not to provide any additional support or damping of the spindle during this process so as not to bias the results. The final diameter achieved was then recorded. Each pass was started at the theoretical critical point and carried both ways to the ends. Where possible, the same dowel used for a given mounting method was re-used for the succeeding methods. It was not possible to do this between Methods 2 and 3, because of the need to add a tenon for Method 3. In this case, two adjacent pieces from the same stick of dowel were used. Three separate runs each were made for Methods 2 – 5, and the results for each method were averaged to produce the results shown in Table 2. The results for Method 5 represent only one example of what can be achieved by adding tension to the spindle. The results would have been different for other spindle lengths, types of wood and amounts of tension. The calculated and experimentally measured relative performances are in fairly good agreement and clearly show the value of the more sophisticated mounting methods. A comment on Methods 4 and 5: Both need a device attached to the live center to grip a tenon on the tail end of the spindle. Furthermore, unless the diameter of the spindle is large, the mass of this device must be small to avoid twisting apart the spindle when starting or stopping. My solution was to make a very light collar out of 1-inch aluminum rod stock by boring one end to accept a standard size of tenon and boring and tapping the other end to screw onto the Oneway live center. I did the boring and tapping directly on my lathe using a Jacobs chuck held in the tailstock. The following figure shows units made for 0.5-inch and 0.75-inch diameter tenons. With this approach, Method 4 potentially would be practical for production turning of spindles, particularly if the alternative is to set up a steady rest. Adding tension for Method 5 complicates the design a bit. For one thing, one needs to keep the tail tenon from pulling out of the sleeve. My approach for purposes of the experiments was to bore a hole crosswise through the sleeve and the tenon and put a pin through it. The other factor is that the live center must be restrained from pulling from the Morse taper of the tailstock and a means is needed to apply tension to the spindle. For purposes of experiment, I made an apparatus consisting of a length of all-thread rod, a thrust bearing, two bearing holders turned from 1-inch aluminum rod, a spring and some miscellaneous hardware. It is shown below. This approach would not be satisfactory for production use, but I can envision other fairly simple designs that might be. The main problem that I see is the need to pin the tenon into the sleeve, although I can envision lightweight clamping arrangements for tenons of a fixed size. Possibly, values of tension much larger than my example would be feasible. The ultimate strength in tension along the grain for most woods seems to range from about 5000 to 20000 psi. (David W. Green, et al, “Mechanical Properties of Wood”, Forest Products Handbook). In my example, the 55 lb force acted on a spindle of final diameter 0.4 inches and produced a stress of about 440 psi, or less than a tenth of the failure load for any of the common woods. As a final comment, I would point out that when turning long delicate finials, where the piece is parted off at the very end, Method 4 would seem to be a much better choice than Method 3, because the finial and the waste piece are fully supported and aligned until the last few fibers are cut.