All 5 books, Edward Tufte paperback $180
All 5 clothbound books, autographed by ET $280
Visual Display of Quantitative Information
Envisioning Information
Visual Explanations
Beautiful Evidence
Seeing With Fresh Eyes
catalog + shopping cart
Edward Tufte e-books
Immediate download to any computer:
Visual and Statistical Thinking $5
The Cognitive Style of Powerpoint $5
Seeing Around + Feynman Diagrams $5
Data Analysis for Politics and Policy $9
catalog + shopping cart
New ET Book
Seeing with Fresh Eyes:
Meaning, Space, Data, Truth
catalog + shopping cart
Analyzing/Presenting Data/Information
All 5 books + 4-hour ET online video course, keyed to the 5 books.
Documentation sketches

For several months I've been working on a very complicated table piece (similar to this
sculpture shown in the first two images here).

In the midst of it all I decided to start removing or revising some of the 11 to 13 elements.
To preserve the progress to date, Andy Conklin, my colleague, made this diagram
showing the location of the various elements in the version immediately prior to
deconstruction. Here is his wonderful documentation, shown here in reverse for beauty

-- Edward Tufte

I like the "perspective-plan" sketch. Anything that can combine two views into one greatly increases information transfer. It is difficult for many people to visualize three-dimensionally from multiple two-dimensional views.

Perhaps in the next book, Professor Tufte can devote a chapter to mechanical drawing in general and construction documents in particular--how to construct something from a set of graphic images and text.

My experience comes from reading construction documents for primarily heavy-highway and industrial construction for over 20 years. Along the way, I have looked at many architectural drawings of buildings, from houses to high rises.

Several concepts come to mind: "de-gridding" data tables, layering and separation, clotted type, reproduction issues, ELIMINATING SHOUTING FROM DRAWINGS, intelligent use of line weights, ridding drawings of obfuscating coding, portraying three dimensions in two dimensions, creative use of sections and views, etc.

For historical perspective, I would look to drawings of early suspension bridges. No computer-aided drafting, no global positioning systems, no total stations with dataloggers, no computerized structural analysis; just a sharp pencil or ink pen, drafting/surveying instruments, and vellum. Even the design of the structure itself came from graphical solutions to load analysis. Today, software guides clients through animated three-dimensional renderings via the Internet.

While detailing concrete form layouts early in my career, a customer wanted forms and ties for a parapet wall. He brought a full scale drawing of a parapet wall drawn with a carpenter's pencil on a sheet of plywood. Inelegant perhaps, yet effective. No scale problems, either. 1 inch equals 1 inch.

Probably the most difficult drawings to read were from a new steel mill. Between the mash-up of numbers, fractions, structure outlines, column lines, extension lines, dimension lines, leaders, asymmetry, changing sections, multiple sheets, etc., it was very difficult to make out the structure itself. Layering and separation would have made reading these drawings far easier, as would metric dimensions, versus Imperial feet-inches-fractions dimensions.

According the the U.S. Census bureau, then Subjects>Construction>Value of Construction Put in Place Statistics; the seasonally adjusted annual rate of total construction is well over one million million dollars per year. The sheer size of the construction industry is reason enough to examine construction drawings and specifications, and how to improve them.

One striking example of how a seemingly small construction detail change led to over 100 fatalities is the Kansas City Hyatt Regency Skywalk collapse. A detailed discussion of how this happened, and the role detail drawings played is part of another post, however.

-- Jon Gross (email)

Dear ET,

The last post describing engineering drawings made me think of other examples of technical drawing methods used as documentation that I have experienced in my professional life.

One of them is "Quickfit??" a brand of borosilicate laboratory glassware that is joined by tapered ground glass joints. All students of practical organic synthesis in chemistry will be familiar with these pieces of apparatus. The really nice thing is that they are very standardized and a simple sketch and/or list of the part numbers is a very accurate documentation of exactly what combination was used. A researcher in another part of the world, or decades later, can build an identical piece of complex glassware and follow exactly the same methods to achieve a synthesis.

An image is shown below of a Soxhlet assembly from the Sigma-Aldrich catalogue (

There is a short Wikipedia article on Quickfit here (


-- Matt R (email)

Fatal Drawing Change—Background

A change in structural steel drawings caused two Kansas City Hyatt Regency Hotel elevated walkways (“skywalks”) to collapse in 1981, resulting in 113 fatalities and 186 injuries. (NIST 1982 page v.) How did this happen?

There were three skywalks that spanned an interior atrium. One skywalk connected the third floor of the hotel to a multi-purpose space (“function block”). A pair of skywalks connected the second and fourth floors of the hotel to the function block. The second-floor skywalk hung from the fourth-floor skywalk, which hung from the roof framing via 6 support rods. Box beams connected the skywalks to the support rods.

The Problem

The fourth-floor hanger rod/box beam connection was the culprit. The design concept was to have continuous support rods from the roof framing to the fourth-floor skywalk box beams to the second- floor skywalk box beams. The steel fabricator saw a 46-foot-long threaded rod, continuously threaded for about 17 feet, with a “captive” nut under the fourth-floor box beam. The fabricator changed this impractical detail to a discontinuous pair of support rods, which effectively doubled the load on the fourth-floor box beam/hanger rod connection. (NIST 1982 page vii. See pages 24 and 28 for plan and section drawings.) The diagram below shows the original design on the left, the actual construction in the center, an end view of the support rod passing through the box beam, and a plan view showing how the channels connect to form the box beam. For scale, the box beam is 8 inches deep, and the support rod is 1¼ inch in diameter:

The original box beam design, with the continuous support rod would have supported the loads acting at the time of the accident; however, the original design would not have met Kansas City building codes (NIST 1982 page vii.) The doubled load caused the fourth-floor skywalk support rods to pull though the weld that joined the two steel channels that comprised the box beams. (NIST 1982 page 77 shows a photograph.)

What Led to the Problem?

When drawing a large object, one can fit the object on the paper using “break” lines, as in the Original Design sketch, above. The break lines allow the drafter to draw the ends at a larger, more readable scale. Unfortunately, the break line graphical tradeoff in the box beam/support rod detail drawings contributed to the collapse. Removing the “redundant” portion of the support rods obscured the reality that the support rods were actually about 46 feet long, with a “captive” nut under the fourth-floor box beam. (NIST 1982 pages 20 and 28.) The box beam/support rod connection as drawn was impractical at best.

“Shop Drawing” Preparation and Approval

At the time of construction, the structural details were conceptual, and it was understood that the steel fabricator would design the actual connections, consistent with code and industry practice. (NIST 2007 page 174.) The steel fabricator then submitted “shop drawings” for approval which described the proposed connections in detail. The owner’s engineer “approved” the shop drawings, after which fabrication and erection proceeded. Obviously, the approval process failed, despite stamps indicating review by the contractor, structural engineer and architect (NIST 1982 page 6.) Engineers are reluctant to “approve” anything in writing, since the very word “approval” connotes liability. Nowadays, engineers usually “approve” shop drawings by stamping them with phrases like “No Exceptions Noted” or “Reviewed.” I doubt if courts waive liability if an engineer stamps a shop drawing “Reviewed” instead of “Approved.”


This is a complicated story; however, some of the solutions are straightforward.

First, redesign the box beam using one or more of the following: place the channels back-to-back, install gusset plates at critical points, place large flat washers at the bottom of the box beam to distribute the load from the support rod, replace the twin channels with stronger structural tubing, etc. NIST 1982 pages 226-227 analyzes the design of the skywalk structural components.

Next, use a coupler nut or sleeve nut to join together two shorter rods into one longer rod. This would have kept the fourth-floor and second-floor skywalk support rods in the original, continuous configuration. The American Institute of Steel Construction’s Manual of Steel Construction describes sleeve nuts on page 4-96 in the sixth edition, dated 1963-65. They aren’t exactly new. Industrial supplier McMaster Carr stocks 1¼ -inch coupler nuts appropriate for joining the skywalk support rods.

The coupler nut, with a modest redesign of the box beam, would have solved the box beam/hanger rod problem—efficiently and inexpensively.

Jon Gross

References available on the Internet:

  1. The Wikipedia article is one place to start:

  2. The National Institute of Standards and Technology (NIST, then known as the National Bureau of Standards) issued its definitive report Investigation of the Kansas City Hyatt Regency Walkways Collapse in May 1982. Be patient, it took over 10 minutes to download this document at 1.3 MBaud:

  3. NIST has another case study issued in February 2007: Best Practices for Reducing the Potential for Progressive Collapse in Buildings, which summarizes the collapse in words starting on page 173:

  4. For another concise article, see: lkway-Collapse.aspx

  5. McMaster Carr is a great industrial supplier. Their Internet site has much useful information at:

-- Jon Gross (email)

Here’s a great anecdote from my experience at a concrete forming equipment manufacturer/supplier. During this time, I saw countless architectural and structural drawings of myriad concrete structures: a swine facility, a sports celebrity’s custom home, high-rise buildings, and literally everything in between.

Walter (rest his soul) was a colorful, big-hearted, old-school salesman. One afternoon, Walter carried in a small sheet of plywood—his customer needed to match a parapet wall. On the plywood scrap, the customer had traced the wall’s outline with a carpenter’s pencil. Walter announced: “I need forms for this wall.”

No dimensions?—1:1 scale—no problem!

Walter also brought in a bill of material written on another piece of plywood. On at least one or two other occasions, other customers brought in a bill of material written on plywood.

No paper? I’ve got some plywood here....

-- Jon Gross (email)

Documentation Sketches: structural details of the Farnsworth House

The Farnsworth House by Mies van der Rohe was constructed between 1949 and 1951. Mies removed every nonessential element of this house—neither applied ornament nor visual clutter defile the Miesian aesthetic. Examine these construction drawings carefully—every line serves a purpose. Attention to structural detail is excruciatingly complete. For example:

Plug welds join the horizontal channel members (15 inches deep by 50 pounds/foot) to the wide-flange (W12) columns. The plug welds are not visible from the front side, as a conventional fillet weld would be.

Coped connections join the corners of the horizontal channel members. The increased surface area of a coped joint makes an intrinsically strong and clean connection—more so than the easier miter joint. After welding, workers ground the welds smooth. From a visual inspection of the actual channel-channel connection is impossible to tell which of the two channels is actually coped.

Prior to painting, workers sandblasted the entire steel framework to remove the mill finish. White epoxy paint completes the elegant finished surface.

During my study of the Farnsworth House, I have seen neither a detail drawing nor written description of the crucial channel and beam connections, much less an isometric drawing like the one below. An early sketch of the beam/column connection shows 6 bolts; however, the as-built configuration obviously lacks bolts and visible welds. The actual size and number of slots in the channel shown in the drawing is reasonable, and the six slots offer approximately 30 inches of total weld length, as fillet welds on either side of the column would. The result is a stunning, "invisible" connection—more elegant than six bolts or two fillet welds.

Thanks to the Farnsworth House site director and our welder for their assistance with this post.

The photograph below shows the portable drafting system that I used to rough out the isometric drawing. I have used this system during and since college. It's a simple clipboard with the clip remounted on birch plywood (for aesthetics) and a strip of hardwood glued along one edge. The raised edge serves three purposes: first, to align small triangles for drafting; second, it protects one edge of papers and third, it acts as a handle. This old friend of mine turns out respectable 8½×11 drawings, especially with a little help from Photoshop....

Computer-aided drafting is great; however, there's no substitute for the ability to sketch an idea on a cocktail napkin or a piece of plywood. It's not always convenient to drag around a laptop and printer. 

Additional images:

-- Jon Gross (email)

I have been thinking about why drawing with pencils on paper is still an essential tool for a visual artist - it cannot be that drawing is the only, or best, way of fixing an image of a scene. John Ruskin already knew this and now with digital imagery it is even more the case.

Below is a statement from the Rhode Island School of Design website describing their foundation courses( I particularly like the phrase "drawing disciplines the eye and brain, tempers judgement, and makes the hand responsive"


"In the drawing studio you will work with the development of skills in perceptual drawing, formal visual principles, and abstract thought. Taught by means of the human figure, landscape, still life, or theme, drawing disciplines the eye and brain, tempers judgement, and makes the hand responsive. You will explore form as it pertains to representation and the organization of surface through line, shape, light, texture, and space. At RISD, drawing is considered the basic tool of all art and design disciplines, reflecting the conviction that this skill "the coordination of eye, hand, and brain" is essential to the way the painter, sculptor, architect, or designer creates."

-- Matt R (email)

Rise of the computer, decline of sketching?

I have received formal, undergraduate training in both pencil-based and computer-aided drafting (CAD), and I have industry experience in both.

One idea became evident to me during my second CAD course: that which is easy and intuitive to do with pencil and paper can be difficult with the computer, and vice versa. For example, measuring irregular areas is easy on the computer and time-consuming by hand. Manipulating a three-dimensional model with a mouse to decide on the optimal view is a snap on the computer but practically impossible by hand in real time. Deciding where to cut a section can be intuitively obvious to a drafter, but the computer has to be told. Often, for a one-time sketch, a drafter can draw more quickly on paper than with the computer. A piece of paper can’t “crash.” I have lost more than one drawing when a cranky computer decided not to cooperate. Overall; however, I must tilt slightly toward the computer.

I'm almost afraid to examine engineering curricula to verify the demise of formal pencil-based drafting instruction. Matt’s Rhode Island School of Design reference above is a refreshing counterexample in the design world, though. I hope that engineering students receive some training in sketching—it is an important skill for on-site communication—be it a scrap of paper or a piece of plywood.

My niece coordinates card design and production for a major greeting card company. For illustrations, she sketches her concept on paper, scans it, and finishes up with Photoshop or Illustrator. I did the same thing to produce the Farnsworth structural system drawing above. As usual, a hybrid method is optimal.

Not many cried for the decline of logarithm tables—like everyone else, I embraced the electronic calculator. Don’t mess with my 0.7 millimeter mechanical pencil, or there’s gonna be trouble!

-- Jon Gross (email)

Dear ET,

Here is an incredible collection of the working drawings of the legendary Japanese artist and print-maker Hokusai.

Orientations Magazine Volume 40 - Number 6 - September 2009

Drawings by Hokusai: Groundbreaking Discoveries Bernard Rousseau

While Hokusai is viewed today as one of Japan's greatest artists, and certainly one of the most prolific, the study of his oeuvre has been hindered by the scarcity of his drawings. Until recently, only ten dating from his apprenticeship years and a few dozen from his later life were known. The recent discovery - in a private European collection - of 24 sheets of preparatory sketches and drawings for illustrations printed between 1807 and 1815 has filled a void. Meticulous examination of these works has shed some light on the period when Hokusai was at the peak of his artistic mastery, and has led to a series of striking discoveries about his working methods.

The website is here =

One of the most interesting aspects of the study is a series of detailed comparisons between the published prints and corresponding preparatory sketches.

"Inquiry into `instructions for the next drawing'.

An unknown feature before the discovery of these drawings, they constitute one of the major discoveries made in the course of this research. It all began when strange peculiarities that repeated in one drawing after another were remarked. The most common was the regularity with which the decoration of numerous items of clothing is carefully indicated, but only once or twice. Another good example is the treatment of the clumps of pine needles: generally, a very few individual clumps are drawn needle by needle with a decisive stroke, while the other clumps are merely indicated by means of an evasive circle.'

-- Matt R (email)

Threads relevant to documentation:

Privacy Policy