The Society of American Period Furniture Makers

Tools and Techniques => Period design and construction => Topic started by: johnah5 on May 09, 2008, 11:00:09 AM

Title: Still at it; Column Orders
Post by: johnah5 on May 09, 2008, 11:00:09 AM
Guys, George, Adam et al,

I have been reading and studying Mack Headley's Tea Table article in FWW.  I have broken down the Ionic Order he uses to numbers and wanted to post them here to get input.  I don't think the width of the leg is correct and would like anyone opinion on this.  I would prefer to have posted a drawing but don't know how to do that on this site.  Opinions on what to use Sketch up etc and how to post would be great also.

Ok, here is what I calculated following the article which uses Ionic Column.

1. Total height that I picked for my table is 30 inches.

2. The Pedestal is 1/5 of the total height.  So 30 divided by 5 equals 6 inches.
    Pedesal = 6 inches.

3. The Entablature is 1/6 of the height less Pedestal.  So 30 minus 6 = 24
    24 divided by 6 = 4  Entablature = 4 inches.

4. The shaft would be the height less the Pedestal and Entablature.
    Shaft = 24inches  (30 minus 6 minus 4)

5. The module is 1/9 of the height of the shaft.  So 24 divided by 9 = 2.222

So, wouldn't that mean that the width of a 30 inch leg would be 2.22 inches?  Also, I noticed that Headley did another calculation that included the Entablature and the shaft to determine the inner and outer limits of the leg.  I am not sure what is going on here and how to understand this calculation.  I took a ruler and measured the drawing in the article and then measured where he labled "module" and tried to determine if the module was 1/9 of the height with the Entablature or what but was not able to determine this.

It seems that a 2.22 inch leg would be too thin to allow a cabriole curve, but I don't have experience with this.  Thoughts anyone?
Title: Re: Still at it; Column Orders
Post by: Mark Arnold on May 10, 2008, 12:46:30 AM
A 2-1/4" blank is small, especially for a chair or a high chest, but may work fine for a tea table. Don't forget that the beauty of the cabriole is the line defined by the diagonal of the blank (at the intersection of the front and side elevations). You unfortunately never see this on an orthographic drawing.

There is a tendency to think of furniture as small-scale, moveable architecture, afterall, much of the nomenclature is the same. But as classical proportions shrink to a personal level, this perfectly imperfect material--wood--gets pushed to its limits. Anyone who has ever made doll furniture knows that a perfectly-scaled piece belongs in a showcase and not a child's hand. I tend to err on the side of caution (especially given the properties of later growth woods) and slightly overbuild certain parts, if I can. Period craftsmen understood the capabilities and limitations of the material they worked yet often pushed the envelope anyway. Consider all of the shield back chairs having short-grain members--many requiring repairs. Today, there are additional considerations caused by the super-sizing of the American physique...

While I've never tried to design verbatim from the orders, there is no doubt that classical architecture has  guided cabinetmakers for centuries. Good proportion is a universal concept--people know it when they see it-- whether or not they can put a name to it. I usually design by eye without regard to any specific order (and perhaps it shows in my work). I've found it more instructive to imitate pieces that are well-designed because the knowledge gained by attempting to repeat a universally-regarded object is transferable to original designs as well.
Title: Re: Still at it; Column Orders
Post by: chamfer on May 10, 2008, 02:58:56 AM
I, too, have found myself confused by the illustrations Mack used to show how he derived various dimensions from the proportions of the Ionic order. However, I think I have sorted it out, and will attempt an explanation.

First, he uses the entire order (pedestal, column & entablature) imposed on the height of the table (say 30 inches for the sake of discussion), to establish the two squares which begin to define the positive and negative spaces (fig. 3). This results in the larger square being 30 inches per side, while the smaller square comes in at 26 inches per side.

In the next illustration (step 3), he considers the column only against the 30 inch table height to determine the outer limits of the knees of the legs. Note, this is different from the thickness of the legs. The resulting module (30" divided by 9, or 3 1/3" - module B) is added to the sides of the smaller square to define the overall width of the table at the knees (and feet). Thus the overall width of the table, at the knees, will be 32 2/3 inches (26" plus 6 2/3").

Finally, in step 4, the dimensions of the legs and the full depth of the skirt are determined by using the column and the entablature of the Ionic order (i.e., minus the pedestal) imposed on the 30 inch height. Thus the skirt (entablature) is 5 inches deep (30" divided by 6), and the legs are 2 3/4"+ thick (30" minus 5" divided by nine - module C).

Having said all that, I find this rather convoluted, confusing, and indirect. Largely, I suspect, because tables may not be an appropriate form upon which to attempt to impose the classical orders. In other words, there just doesn't seem to be enough "facade" there upon which to develop the "vocabulary" and "grammar" of the classical orders. Though I certainly make no claim of expertise regarding the classical orders, so  my comments are mostly an effort to further conversation on this topic.

Don McConnell
Eureka Springs, AR
Title: Re: Still at it; Column Orders
Post by: walkerg on May 11, 2008, 11:37:01 AM
I do agree that the overall tea table form is blocked in using the proportions from a classic order. The top and stretcher are essentially an entablature and the height of those elements would be arrived at using very simple divisions found in a classic order. The cabriole leg is a whole other matter. First some background on the form itself. Regardless of whether the leg ends in a claw and ball or pad foot this evolved from a classical form known as a scrolled bracket or modillion. This is an “S” shaped form with two opposing volutes, one volute always larger than it’s opposite. We see it in Greek architecture as far back as 5th century B.C. but the form matures under Roman architects who used it as a bracing element to support the entablature on a Corinthian classic order. Over the centuries the scrolled bracket was applied as a supporting brace in a variety of applications and also as a keystone in a Roman arch. The form can be applied horizontally, such as in the application in a Corinthian order, or it can be oriented vertically. Renaissance architects used scrolled brackets vertically as legs to support alters and tables. Usually the larger volute was oriented at the top. This would correspond to what we term as a knee on a cabriole leg. Often the lower volute was replaced with an animal foot such as a lion’s paw or a hoof. Frequently the foot rested on a plinth, a small rectangular or square block. You might not readily make the connection between these table legs and the period ball and claw primarily due to the medium. The renaissance examples are typically modeled in stone with much bolder curves and heavier proportions than the same form in wood. The same difference in proportions can be seen in turned parts like spindles to support a stair banister. The stone version (that inspires the forms we see in wood) is much bolder and stouter than its wood counterpart. A good treatment on this can be found in the book by Robert Adam - Classical Architecture, A comprehensive handbook of Classical Style, Abrams, 1990.
Concerning the layout of a cabriole leg, I have not seen anything in any of the period literature I have read. Although, if you think of it as a stretched out and slender scrolled bracket, most of the design books of the day do explain how to lay that out. We can only speculate, but my guess is they divided the height into equal units, perhaps 16 or 20 to establish a module. That module was then used to define all the major transitions and boundaries for the form using simple proportions. I would love to get some profiles from different cabriole legs known to come from the same shop such as Chapin in Connecticut. I’m fairly certain with some dividers and patience one could unlock a proportional scheme that would allow the form to adapt from a dining chair to a highboy to a dressing table. I’ve often wondered if the master of a shop would work out and refine a form for a cabriole leg, make a wooden pattern so his workman could apply it, but as a last step, plane off his layout lines to prevent the real knowledge from walking out of his shop.

George Walker
Title: Re: Still at it; Column Orders
Post by: johnah5 on May 11, 2008, 08:11:44 PM

I think your comment about planing off the layout lines to not show the secret may be the key here.  Do you think I am trying to fit a square peg in a round hole?  Maybe the individual used the column orders/Fibonacci numbers and golden ratio for some things and not others. The column orders can give a staring point of say 1/5 or 1/6 of the total size for different parts to a piece of furniture.  Then one could use the golden ratio to determine where the waist molding or bottom to top division would be.  Then they adjusted parts by eye?  Just a WAG.  I am trying to find guidelines to follow to design pieces I want to build.  Do you think there was a comprehensive system that the Master used, pulling out different parts of column orders and other methods to come up with widths/heights and depths?

This issue is fascinating to me but I am really fumbling through this.  I appreciate all the posts.

Best regards
Title: Re: Still at it; Column Orders
Post by: Wiley Horne on May 11, 2008, 08:56:02 PM

Please check the arithmetic.  If I follow your numbers, I get 20" for the shaft (30" - 6" -4"), which would also affect the downstream numbers from there.


EDIT:  Never mind.   Just a typo in the original post.  The 2.22" you got is 20/9.  Sorry
Title: Re: Still at it; Column Orders
Post by: johnah5 on May 14, 2008, 09:52:38 PM
I have been braining this a lot and was able to email converse with Don McConnell a bit.  I want to put forward a theory that may be obvious to others.  It seems that the column orders, golden mean, fibonacci numbers etc, were all used to make furniture at the same time.  It seems to me that there is not a rigid system used to design members of a piece and that it may be more likely that the cabinet maker used say, 1/5 or 1/6 sizes from columns and other ratios for other parts of the piece. 

In a recent blog from Chris Schwarz talked about a piece he designed for the magazine and he did it by eye.  At the end of the design he calculated a number of the measurements and saw that the golden mean, the column orders etc were present.  He didn't start with these ratios or guidelines, it seems they naturally grew from adjustments to the dimensions he made by looking at the piece.  The other salient point was that there was no overall guiding order.  In other words it was not all column orders or other method.  The piece contained a mix of these orders not a rigid set of one of them.  Are we making too much of these systems.  Maybe the cabinet makers used a little of this and a little of that and adjusted per eye to get the "look" right.


Title: Re: Still at it; Column Orders
Post by: mbholden on May 15, 2008, 09:54:50 AM
I would think it would go more like this:

I want an addition to my house, so I draw up a sketch, and go to the architect with it.
The architect takes the lines that represent the walls, and fills them out with 2/4's drywall, sheathing, panelling, siding, etc. and brings the drawing back for approval.

The same with furniture design:
Draw it up freehand, refine till it looks good, then finish by using known esthetically pleasing ratios and arrangements. Then apply the ornamentation that is indicative of my chosen style.

BTW, the book that Chris Schwarz refers to in the design article, The Old Way Of Seeing, is quite good. I would recommend it highly, as it deals not only with design, but how to get there. Caveat: I am only four chapters in. (I ordered it when I read the article)

Title: Re: Still at it; Column Orders
Post by: johnah5 on May 15, 2008, 10:35:58 PM
Thanks for the suggestion.

I spoke to Chris about this issue and he is unaware of a system of ratios based on column orders etc.

I would guess that there are a lot smarter minds who have pondered this issue.  I am hoping to learn from them instead of trying to arrive at the conclusion on my own.  Sort of a community approach to trying to figure this out. 

I am hoping this thread will generate more thought and discussion on this topic.  I am not sure if it would be better to post a new thread so that people would see it.

Title: Re: Still at it; Column Orders
Post by: chamfer on May 18, 2008, 11:25:25 AM
Hi John et al,

There is no textual evidence, that I am aware of, that the golden ratio or fibonacci series were ever used in furniture and architectural design prior to the latter part of the 19th century. There were some 20th century efforts to use the golden ratio by a couple of leading architects, but, in practice, they found they had to modify the ratio (thus negating any sense of it being some sort of "special" ratio), and, even at that, found it to be of very limited utility. It seems to me that a more useful approach for indicating the proportion of some aspect of a furniture design would be to use simple ratios such as 3:5 or 5:8, just to name a couple of examples.

By contrast, virtually every 18th and early 19th century text on architecture and furniture contains information on the classical orders. For this reason, there exists at least some rationale for thinking about the ways in which the classical orders may have been used, historically, in period furniture design.  Though, at the same time, I think that this needs to be approached with some caution. The fact that none of the authors (as far as I know) specifically outlined a systematic method for applying the classical orders when designing furniture suggests that no such approach was universally known/employed.

Further, as I already hinted in an earlier message, some forms of furniture do not seem to lend themselves to this approach. For example, in looking at my copy of Chippendale's _Director_, I'm always struck by the fact that illustrations of the various chair designs seem to bear no relationship to the just preceding illustrations of the classical orders, mouldings, volutes, etc.

On the other hand, some pieces of period casework seem to have been largely informed by the classical orders (one example which come to mind is an early 18th century Swiss wardrobe shown in Greber's _The History of the Woodworking Plane_, p. 169), so it seems worthwhile to explore some of the possibilities.

George's mention, in a parallel thread, of a rough rule of thumb for a crown, or cornice, moulding of a piece of furniture to be about 1/18th of the total height reminded me that Batty Langley has a similar method for determining the height of room cornices in _The Builder's Jewel_ - page 28 and Plate LXXVI. Though these formulas are for architectural work, I think they suggest the possibility that furniture designers/builders may also have used similar shortcuts for indicating the sizes and proportions of specific aspects, such as cornice mouldings, of pieces they were working on.

Langley's formulas, though, suggest a wider range of possibilities than the single proportion of 1/18th of the overall height of a piece of furniture. He further extends the possibilities in his treatment of the same topic in _The Builder's Jewel_. I've explored that a bit, and will include that in a separate post.

Don McConnell
Eureka Springs, AR
Title: Re: Still at it; Column Orders
Post by: johnah5 on May 18, 2008, 01:32:37 PM

Thanks, as always a great post.  Will people see this since it is buried down at the bottom of this thread?  I would like to read more opinions and thoughts on this topic.  Should you cut and paste this as a new thread?


John H
Title: Re: Still at it; Column Orders
Post by: chamfer on May 18, 2008, 03:27:02 PM
John et al,

Per my previous post, I wanted to follow up on Langley's short-cut formulas for determining the height of room cornices. The relevant text from _The Builder's Work-Bench_ is in the pdf file below (tried typing it into the message window, but there is some glitch between my computer and the forum server, and the text keeps getting mildly mangled). As can be seen, they extend the possible ratios from the three given in _The Builder's Jewel_ to nine, depending on whether one is considering the cornice to be a part of the entablature atop the column alone, atop the column plus a plinth, etc.

In attempting to envision the relationships between the various resulting ratios, I decided to make up a quick sketch to allow for direct visual comparison. That is the gif file below. I also included the 1:18 ratio suggested by George. The letters in the sketch correspond to those in brackets I added to Langley's text in the pdf file. Incidentally, this wasn't as complicated to draw as it might seem. I found it to be fairly straight-forward, using the various scales on a triangular drawing rule.

A few thoughts/observations seem to arise from Langley's text. One, is that the 1:18 ratio falls near one end of the spectrum of possibilities. Another, is that Langley, at least, envisioned a good deal of flexibility in deciding the height of room cornices based on the classical orders.

What Langley doesn't provide, however, is any idea as to what factors might go into deciding between the various options. But, I think it likely that the choice would largely be driven by a couple of factors. One obvious factor would be which classical order may already have been in play in the room. Another would be whether the intent was to project strength and authority (Tuscan, e.g.) or restraint and refinement (Corinthian, e.g.). Playing with these possibilities might be useful in designing furniture as well.

I think we can draw one other inference from Langley's illustration in _The Builder's Jewel_. Across the three possibilities, he indicates a chair/dado rail height of two feet ten inches. Since there is no indication that this is scaled to any given room height, it seems this rail height was fairly standardized (at least in Langley's mind). If he were rigidly scaling things to the relevant classical order, we would expect this height to vary according to the height of the room. I suspect there were questions of practicality and human scale which led to the standardization. The same kinds of considerations are even more critical when it comes to designing furniture.

Needless to say, I'd be interested in other people's thoughts on these issues.

Don McConnell
Eureka Springs, AR
Title: Re: Still at it; Column Orders
Post by: deanj on May 19, 2008, 11:16:29 AM
Don, John et al,

I've pondered the column orders as well.  As Don notes there are references in nearly all of the existing texts from the period.  I've also gone back and read Vitruvius  (Ten Books on Architecture ) and Palladio (Four books of Architecture) to try to learn from the texts likely to have been used by the cabinetmakers.  I think Palladio has more "rules" laid out, but again for the construction of buildings, sizes of rooms, etc.  One tidbit I'd like to spend more time looking at is the "rules" for stacking the orders.  The first floor of a building has a certain order, the floor above it has an order "higher" in the progression.  [Tuscon (low) -> Corinthian/Composite (Highest)]  Do chests on chest use this rule?  High chests?

Having spent some time, not nearly as much as many others, I have a sneaking feeling the orders are more like scales and keys in music.  A common framework by which to structure something.   We know a musical key can evoke a mood, just like a classical order conveys power or grace.  Given the key we can use the proportions within that key (order) just like the particular notes/chords within a key.   But, there may not be a solid set of rules.  Given a key think of all the styles of music which can be created!  Each style has its own set of rules, simple rock may follow a chord progression of 1, 6, 4, 5, 1  and be written in 4/4. Swing may be in 4/4 but with a much different chord progressions, and so on.  So, even with a common framework the way you apply the "rules" varies greatly.   

Perhaps I need to start practicing my scales...

Title: Re: Still at it; Column Orders
Post by: msiemsen on May 19, 2008, 01:18:18 PM
 Since Dean's post referred us to musical keys I wondered if the older Modes would have any correlation. Being a musical illiterate I offer them to you to peruse.
    Ionian, Dorian, Phrygian, Lydian, Mixolydian, Aeolian, Locrian

Title: Don's post
Post by: johnah5 on May 19, 2008, 02:03:36 PM
Hi Don,

Thanks for the post and for taking the time to make the drawing and post it.

Couple of quick questions/thoughts.  In looking at your chart there is not a lot of difference between the different ratios save 3:35.  The rest are quite close and I would choose a cornice based on what looked good to my eye. 

Given that the differences in size of the conices in your chart, is not that drastic, do you think the chart shows that we are back to the individual using his personal preference to make the decision and not a specific set of rules?

Can someone recomend a way for me to look at pieces (either via web or a book) so I can take a ruler to them and calculate some of these measurements?

John H

Title: Re: Still at it; Column Orders
Post by: chamfer on May 19, 2008, 10:49:20 PM
John et al,

At the size of my sketch of cornices (as downsized for my previous post), I can see where you might think the variations are not all that great. In fact, I think they are quite significant, and meant to say something about it in that post. While it's obvious that the difference between the two extremes is huge, there is also a significant range within each order. Though the overalp between the orders tends to obscure this fact.

I thought it might be helpful to quantify these ratios in order to illustrate. If we were to assume a piece of furniture seven feet tall, these are the cornice heights (rounded to the nearest 1/32") for the various ratios:

1:18       -     4  21/32"

A. 3:35   -     7  3/16"
B. 3:39   -     6  15/32"    Tuscan
C. 3:43 3/4 - 5  3/4"

D. 3:40   -     6  5/16"
E. 3:44   -     5  23/32"     Doric
F. 3:50    -    5  1/32"

G. 1:15   -     5  19/32"
H. 4:66   -     5  3/32"       Ionic, Corinthian & Composite
I. 4:75    -     4  1/2"

Looking at the Ionic order, for example, it might seem that a difference of a little over an inch between G and I isn't all that significant. However, if you were to mock up the same cornice moulding at the two different sizes and offer them up to an appropriately sized piece of furniture, I think you might be shocked by how much of an impact the difference makes.

And, yes, I think it does come down to a matter of judgment, based on the overall size of the piece (including the width), other existing proportions, the surroundings, etc. Though, I think that using such ratios as a starting point, first on paper, then, possibly, in mock-ups, could be a very good way to begin the process of figuring things out.

The matter of scale always makes it difficult to tell such things from drawings or photographs. For this reason, any serious study of proportions of existing furniture is best done through accurate measurments of the pieces themselves.

Don McConnell
Eureka Springs, AR
Title: Re: Still at it; Column Orders
Post by: walkerg on May 20, 2008, 08:07:36 AM
Concerning Dean's question on stacking the orders. As with most areas of classical design there are few specific rules but there is a principle at work. Typically on a vertical plane one should avoid repetition. The Romans answer to this problem when they began building mutlti story buildings was to superimpose the orders on each succesive floor so that each floor is slightly shorter than the one below it. The column diameter is smaller at the capital so by starting the next floor with the smaller diameter it makes the next story shorter. They usually used a Doric on the bottom progressing up through an Ionic and Corinthian. If you would use only one order for all floors the succesive stories would diminish too dramatically. We use the principle in graduated drawers to break up the monotonous repetition of stacked elements. Don't bother with stacking classic orders though to arive at a method, just understand the principle behind it and use the width of the drawer divider to reduce each opening to achieve the effect.
It's tempting to want to look for rules that will give us step by step instructions on design. Sort of a paint by numbers approach. Instead we can learn more if we look for the principles underlying, and study good models to see how those principles can be applied. For example, a Roman Corinthian classic order offers a great lesson in laying out decorative carving. It is often the most elaborativley carved. Notice though, that the is carving intersperced with plane surfaces. In fact the plane surfaces are larger than the carved surfaces. That's very important, too much ornament can make a design look like a briar patch and not allow the eye to focus and enjoy.

George Walker
Title: Re: Still at it; Column Orders
Post by: walkerg on May 20, 2008, 08:28:49 PM
More thoughts on sizing a cornice. My earlier references to using a classic order to locate and size a chair rail on a wall, or using an 1/18th ratio for sizing a cornice are pretty standard examples of explaining how a classic order can be applied. I wasn't my intent to define hard and fast rules for either case. Its very important to begin to be able to vissualize the inumerable ways the orders can be used. Once you begin to visualize the basic proportional principles you can more readily interpret what you are seeing. Back to the cornice example, for a piece of case furniture it's inspired by a classic order but typically in a much modified form. Designers understood well that as elements are scaled larger or smaller, adjustments must be made. For example, a colunm over 30 feet high needs less taper because of the effects of perspective at the greater height. For corninces, most of the design books offered scaled down versions for door and window treatments more appropriate to the smaller up close application. My thought is that most moldings for furniture were scaled down versions of the door and window treatments. Less projection, often even eliminating a freize or bed mold etc. That being said, the underlying principle is that the cornice should have a proportional link to the case below it. As always classical design is about studying good models and using them to train our eye and guide us. The classic order being the original model and the key to understanding all that is inspired by it.

George Walker
Title: Re: Still at it; Column Orders
Post by: johnah5 on May 21, 2008, 10:43:52 PM
I think you may have brought us to a conclusion on this topic.  I spoke to a very accomplished cabinet maker who has made numerous period pieces.  When I asked about column orders his response was that he does not use them, instead he is making things based on the following; a historical piece, a piece that has to look right in a given room (ceiling height etc taken into consideration) and whatever size router bit or cutter he has on hand.  He acknowledged column orders but just was not designing/modifying based on them.

I am beginning to conclude that the orders were and inspiration as George has said but other factors outweighed any rigid set of rules based on an order, such as the tools on hand and possibly the stock on hand.  So in the end, say 1/18th, 1/5 or 1/6 may only be a great place to start and not an integral clue to a hidden system.

Just another thought...


Title: Re: Still at it; Column Orders
Post by: walkerg on May 22, 2008, 04:25:17 PM

I'm not so sure we have concluded this topic as much as just begun it. Classical design is a tradition that goes back 2600 years and is our direct link with the period artisans that SAPFM is devoted to keeping alive. Several times in history the in depth knoweledge of that design language has been lost only to resurface in unlikely places (18th century American colonies for example). It's true that very few modern artisans now have a working knoweledge of the classical approach that was common to our forebears, that does not in any way make it irelevent. Thirty years ago hand tool makers had pretty well died out and it was very difficult to find good information about how to tune a hand plane, let alone use it. Now we have a tremendous resurgence of both tool makers and users who are refilling the gaps in our lost hand tool heratage. If we had all stuck to router bits and sandpaper back then, the craft would have lost a great deal. Think of the study of the classic orders on par with an artist learning the basics of the color wheel, and the effects of light and shadow and perspective. True, you can paint without that knowledge, a good copiest may even be able to produce a marvelous rendition of a master work. But think how much is missed. Our forebears routinely studied and copied the great masterpieces they had access to, that is part of the classical tradition. The focus though was always so they could understand the original master. This is still relevant and can bring tremendous vigor into your woodworking experiance if you let it. A little bit like learning to use a good finish plane. The learning curve can be a bit steep at first, but the rewards more than compensate.

George Walker
Title: Re: Still at it; Column Orders
Post by: johnah5 on May 22, 2008, 11:17:38 PM
Thanks George,

Good point.  I could use some direction however.  There is too much old material out there for me to start on.  I am going to try to locate 18th century furniture here in Indianapolis to look at, and hopefully take measurements.  Would appreciate any direction on where to continue so I can attempt to absorb some of these mysteries.

Title: Re: Still at it; Column Orders
Post by: Adam Cherubini on May 23, 2008, 11:53:12 AM
If you walk around the furniture wing of an art museum like Philly, you quickly see that similar pieces have all the variety of the human body.  But like the human body, the differences are subtle variations of the same basic shape. 

Leonardo Da Vinci started his work with a few basic proportion rules.  He split the body into 6 or 8 equal parts vertically.  He superimposed a grid system based on his dissections, direct observations, and observations of ancient sculptures.  These represented a baseline from which his self portrait or the mona lisa was done.   

Philadelphia has several of Thomas Eakins drawings that he did as preparation for his paintings.  The drawings are carefully, almost draftsmansly done, indicating fader points, horizon lines, reflections etc.  Do the finished paintings (which are adjacent to the studies) perfectly match the drawing? Of course not.

The basic proportions establish the composition of the work.  Anybody who thinks they can just wing it is....well...let's put it this way John; One's work will improve if one uses the classical proportions.  I don't think you have to hold to the dimensions to the third decimal place.  And it's just silly to think that because we don't see that level of accuracy on 18th c furniture that no rules were used. 

For my part, until I get to be a better artist than Da Vinci, I'm going to stick with some sort of classical proportioning scheme.