Lacquer vs Silver

[Dan Schultz]

If we had Rick Denney calculate the approximate area of that horn, we could get an idea of how thin the plating on it really is - and my bet is that the thickness of the silver plate is barely thicker than that of laquer.

The thickness of silver is but a fraction of the thickness of lacquer.... barely more that raw brass. As far as horns with different surface treatments playing differently goes... well, there are a heck of a lot more things that affect the horn more than the finish.

[Joe Baker]

If we had Rick Denney calculate the approximate area of that horn, we could get an idea of how thin the plating on it really is - and my bet is that the thickness of the silver plate is barely thicker than that of laquer.

'Ye Olde TubeNet' featured just such a calculation:

http://www.chisham.com/tips/bbs/jun2003 ... 39212.html

Rick gave the number; my reply to Rick's post shows how I arrived at the same figure when doing my own calculation (farther down the thread, Rick reveals that his method was very similar to my own, though he probably used calculus or taylor series polynomials to determine the surface of the bell whereas I used plane geometry to determine the surface of a simple cone).

Anyway, as to the thickness, you lose your bet. Working backwards (assuming a thickness of .0004 inches and calculating the weight) we arrived at 12 oz of silver, which a couple of platers told us was probably twice the correct amount of silver. So the average thickness of silver plate is closer to .0002 inches! The thinnest lacquers are MUCH thicker than the thickest plating.
___________________________________
Joe Baker, who nodded in agreement when reading Sean's comment about comfortable shoes.

[Rick Denney]

Rick gave the number; my reply to Rick's post shows how I arrived at the same figure when doing my own calculation (farther down the thread, Rick reveals that his method was very similar to my own, though he probably used calculus or taylor series polynomials to determine the surface of the bell whereas I used plane geometry to determine the surface of a simple cone).

You must be confusing me with Chuck.

Being a genius means calculating the correct answer by the most efficient means. That gives me two opportunities, when only one is usually enough, to miss the mark.

Rick "who recalls Dan Oberloh weighing the silver itself during a plating process, but not recalling what his answer was" Denney

[Rick Denney]

Sounds like a good question for Roger Lewis, if he's willing to dig out and play a bunch of identical models in different finishes from WWBW's inventory.

There is a way to do this, but it takes much more than what you suggest. Roger Lewis, like everyone, has biases he cannot ignore even if he wanted to.

Here's an outline for some ambitious grad student wanting to do a thesis.

Each population of the same model of instrument will show sample variation. But that sample variation should vary randomly around a mean, or average. You can represent that variation with a bell curve showing the probability that a given instrument will vary by a particular amount from the mean.

The trick is to test enough of a sample of lacquered instruments to draw that bell curve. Then, perform the same test with a large sample of silver instruments. That will also produce a bell curve.

The bell curves are approximations, based on the assumption that the sample is smaller than the whole population of tubas of that model. You'll end up with a fuzzy scatter from one sample sprayed on top of the fuzzy scatter for the other sample.

Enter statistical analysis. Statistics are the tool scientists use to see if there is a real difference hiding in the noise of those two samples, or if any apparent difference is just more of the same random noise that is part of normal sample variation.

Obviously, the smaller the difference in the samples, the bigger the samples have to be to show that difference.

Not too many tuba grad students know statistics, so this is not a thesis I expect to see any time soon. But the statistics are the easy part (to anyone with a couple of college-level applied statistics classes and access to statistical analysis software).

The tricky part is figuring out what test to conduct to measure each population.

The test can be subjective, as long as the judge cannot see the instrument and therefore cannot be biased by what he sees. This is called blind testing. The test might be as simple as rating the "brightness" of the sound on a scale of 1 to 10, though the more subjective the test, the noisier will be the resulting scores, and the bigger samples you'll need to prove a real effect. A large panel of judges will filter out a lot of that noise.

It would be best if the performer also did not know the finish on the instrument. To achieve this, I would suggest that the performer play a memorized suite of test music blindfolded, with the instruments placed in random order and handed to the player by an assistant, using the same mouthpiece for every test. This makes the test a double-blind test. I'm assuming that a silver instrument doesn't feel any different than a lacquered instrument, and I'll bet that's the case, but for proper rigor it would have to be tested separately.

You could have the player record his test judgments just as does the judge, to see if there is an effect that is felt uncorrelated to what is heard out front. Playing through all the instruments multiple times (but in random order on each new cycle) would reduce the noise of the player's score.

If the two samples comprised ten instruments each, and the scatter of the data completely overlapped with no statistically varifiable difference in the two samples, that might persuade people that the effect is too small to believe in. Eight of each might be enough. But one of each is utterly unpersuasive--it is impossible to quantify the sample variation so that the effect can be isolated from it.

Rick "who would bet on there being no measureable effect" Denney

[Mudman]

Rick has some good ideas on testing the "lacquer vs silver" question.

At the University of Iowa, there was a study on bell materials and their effect on tone quality. The study was conducted using a french horn with four different screw-rim bells. If I remember correctly two bells were laquered, and two unlacquered; and some of the bells had been annealed. Bells were tested in a double blind setting (listener and performer were unaware of bell construction). Listeners rated tone color from darker to brighter along with possibly one or two additional criteria.

Problems that arose included the length of time between each playing sample. Bells had to be changed, resulting in at least 20 seconds of delay between each sample. Add to this the clanking sound of metal bells being changed, making it difficult for the listener to remember what the first tone really sounded like.

The dissertation did suggest that bell composition did play a noticeable role in tone color.

A better test would use some kind of synthetic lips that could quickly and silently be switched from one tuba to the next. The lips could be clamped on to a mouthpiece insuring consistent tone production. Each tuba would need to be played in exactly the same position to accurately compare subtleties in tone, necessitating some kind of wheeled rack for the instruments. (Aiming the bell one or two inches off would probably create different results.)

Another study was done at the ITF in Denton TX. Trombones were played behind a screen by famous players who were blindfolded and who were wearing padded gloves. The problem with this survey was that not a large enough sample was used. It did suggest that the "hottest" horn (Edwards) was not the easiest to play, nor did it sound the best.

Ahhh, academic pursuits. I'm out of here, time to practice.

[Rick Denney]

At the University of Iowa, there was a study on bell materials and their effect on tone quality. The study was conducted using a french horn with four different screw-rim bells. If I remember correctly two bells were laquered, and two unlacquered; and some of the bells had been annealed. Bells were tested in a double blind setting (listener and performer were unaware of bell construction). Listeners rated tone color from darker to brighter along with possibly one or two additional criteria.

...

A better test would use some kind of synthetic lips that could quickly and silently be switched from one tuba to the next. The lips could be clamped on to a mouthpiece insuring consistent tone production. Each tuba would need to be played in exactly the same position to accurately compare subtleties in tone, necessitating some kind of wheeled rack for the instruments. (Aiming the bell one or two inches off would probably create different results.)

...

Another study was done at the ITF in Denton TX. Trombones were played behind a screen by famous players who were blindfolded and who were wearing padded gloves. The problem with this survey was that not a large enough sample was used. It did suggest that the "hottest" horn (Edwards) was not the easiest to play, nor did it sound the best.

The synthetic exciter exists, but I'd rather filter out the player effects. It would be more fun.

I hadn't thought of using gloves--that solves a problem.

With the appropriate assistants, the delay between instruments could be as little as a few seconds. All the activity should happen behind a good-sized screen to prevent the visual distraction, the pollution of the blind test, and any minor differences in position. And recycling through the samples, in random order, would control for the effects of position within the list.

Different tests could be conducted. For example, one round of tests might be playing a single, loud note, such as a C on the staff, at FF. That would make it easier to remember the sound from one instrument to the next.

For example, if the player played a single, loud staff C, held for four full counts, and repeated every six seconds, two sample sets of ten instruments each would only take 3 minutes and 20 seconds. Five repititions, with the instruments in random order, would take 20 minutes. Then, you'd have a break to "clear the palate".

The horn test at Iowa used samples that were too small. When one studies statistics, one learns that "n-1" appears in most statistical formulas, meaning that a sample of one renders the formula meaningless. Even with a glow-in-the-dark effect, you'd need a sample of at least three of each that were nominally identical. The more subtle the effect, the bigger the sample needs to be to reveal it.

The performer needs to be good enough to be able to play the same lick over and over again consistently, but I don't think the player would have to be world-class. We would be looking at instrument effects, not player effects. IN fact, if the player was too good, he might automatically adjust his sound to smooth over any differences that he senses.

I think the whole thing could be done at a conference (perhaps the Army Conference in some future year), with 100 judges taken from the attendees. It would take 10 or so folks to handle the instruments, and need perhaps a day to set up the randomization and labeling. You'd line up the instruments in a row, with their order for each repitition being printed on a label. The instruments and the performer would be behind a screen big enough to hide it all. The performer would be blindfolded and gloved. The data anlysis could be done in a day, with some advance preparation, and the results presented at the same conference. Roger's role could be as the performer, and helping persuade Miraphone to provide the instruments for the test.

Rick "who would volunteer to help" Denney

[Mudman]

I think the whole thing could be done at a conference (perhaps the Army Conference in some future year), with 100 judges taken from the attendees. It would take 10 or so folks to handle the instruments, and need perhaps a day to set up the randomization and labeling. You'd line up the instruments in a row, with their order for each repitition being printed on a label. The instruments and the performer would be behind a screen big enough to hide it all. The performer would be blindfolded and gloved. The data anlysis could be done in a day, with some advance preparation, and the results presented at the same conference. Roger's role could be as the performer, and helping persuade Miraphone to provide the instruments for the test.

Rick "who would volunteer to help" Denney

Yeah, I'd be in to helping/organizing as well. Just don't put me in charge of lifting all of those expensive horns. I couldn't handle the stress

Any doctoral students want to take a shot at this for a dissertation project? This would be a sure thing.

[IkeH]

I've had an older Piggy CC for about a year and after getting it in good physical/playing condition, I stripped it since it had less than 50% of the original lacquer. With the same horn, player, mpc, practise room, I noticed a TAD bit more brightness that wasn't there before. I know this wasn't much of a scientific test, but strictly a subjective unplanned reaction to the change in the horn. I don't remember a change in the response, although there was definitely one after doing the dent work, resoldering, etc. After having horns in every sort of finish, though, it just comes down to whichever one works the best, however the finish does or doesn't affect it. You get more comments on silver at Tubaxmas, though.