By Mark Shepard
From Restoration Agriculture

In attempting to design a Keyline system on this property I discovered all kinds of things that just didn’t make sense. It all started with the basic vocabulary describing land shape. It was fairly easy to define the “hills” on this farm. There were four low “knuckles” rising out of the general lay of the landscape, the highest in elevation abutting property owned by none other than Bud and Dee Hill. The Hills on the hill. It was fairly simple also to find the primary valleys. They were the valleys that cut into the sides of the main ridge where water first began to collect and flow in flash-flood events. But that is where a dogmatic adherence to the Yeomans’s plan began to unravel. For one thing, there were some primary valleys on the farm that had no clear keypoint. There were other primary valleys that appeared to have multiple keypoints cascading down the primary valley like a series of pools in a mountain stream. Which one was the “true” keypoint? Where do I start?


The next puzzler for me was the fact that yes, the main ridge was cut into by primary valleys, but in our case the primary valleys didn’t feed a “main valley” but joined another primary valley to form a secondary valley. The secondary valley joined with another to form a third that joined a fourth, then finally a fifth (with the named Camp Creek in the bottom) before it reached what would have qualified as a main valley in Yeomans’s terminology. But even that continued on… What I considered the main valley—Camp Creek—joined with the east branch of the Kickapoo, which joined the west branch of the Kickapoo, which joined the Wisconsin River, which joined the Mississippi before returning to the ocean. Nine valleys?

What I had discovered was that the Yeomans terminology completely failed to describe the landscape that I was working in. What I had discovered on the ground (and not from a book, a satellite, or a GPS unit) was what is known as the Strahler, or Horton-Strahler, stream order classification system used by ecologists and hydrologists worldwide. In the Strahler system, a Yeomans “primary valley” is called a first-order stream. When two of these first-order streams come together, they form a second-order stream. If the second-order stream is fed by primary valleys only, it still remains a second-order stream, but when a second-order stream meets with another second-order stream, they become a third-order stream. This combining of stream orders continues until, as in the case of the Mississippi River, you get a tenthorder stream.

Horton-Strahler stream order classification system
The Horton-Strahler stream order classification system. “Primary valleys” in Keyline design vocabulary are “first order streams” according to the rest of the world.

The majority of streams in the world have a stream order of three or less, and it is within that context that the Yeomans plan was developed. I was attempting to take a simple water management system developed in a geographically simple landscape and apply it to one of the more complex watersheds in all of North America. As a matter of fact, I was attempting to apply it to the most complex watershed on the planet! Although 3+5=8 is beautiful, accurate and true and perfect every time, the mathematics that second grader uses to solve that problem are not adequate for solving.

Why did it matter that the Mississippi River watershed is so complex? First, the complexity of the water system made it difficult to find the keypoint. It appeared to me that this farm’s primary valleys had several keypoints, but according to page 13 of Water for Every Farm: “ONLY A PRIMARY VALLEY HAS A KEYLINE” (caps original). If only a primary valley has a Keyline according to Yeomans, then it follows that only a primary valley has a keypoint from which it is derived. Simple! But wait a minute…. A few pages later (page 32), Yeomans writes:

The Secondary Valley

On occasions a series of primary valleys on the one side of a main ridge will join up with a larger valley, which does not contain a channeled water course in the bottom of it. Such is named a “Secondary” valley, and it will have at its commencement its own keypoint and Keyline.

Now I was really confused. First, Yeoman’s says that only a primary valley has a keypoint, yet at least one primary valley on this farm had what appeared to be several keypoints. Then I read that not only do primary valleys have keypoints but secondary valleys do as well? How can both be true?

hill and field
Although the principles of Keyline geometry are simple, most land forms are NOT! This photo shows the
infamous “starting point” on New Forest Farm where multiple, unclear, “apparent keypoints” were maybe somewhat visible.

I staked out and flagged many of the other apparent keypoints to see how the geometry would work, and none of them really did. Again I was confused. It turns out that while actually attempting to design a system on the ground, I had discovered something Yeomans only once barely mentions. On page 47 of Water for Every Farm, he writes, “some steep primary valleys cannot be cultivated as described, because the shape of the valley contours may make turns in the valley floor impossible.”

Now, not only was I attempting to use first-grade math to send a spacecraft to Jupiter, I was attempting to take a simple water management technique designed in simple land forms, and apply it to a complex landscape, and not only did I have primary valleys, secondary valleys, and third-order valleys to deal with, I had multiple keypoints and contradictory information about them. I was setting out to do what the master himself claimed was “impossible.”
Yeomans’s recommendation on how to deal with such tight primary valleys did come to inform my designs later on, however. Later, when describing tight primary valleys, he describes in one brief sentence a technique which I have come to learn is as revolutionary as the Keyline pattern cultivation itself. “These valleys are most suitably worked in a herringbone pattern with a tractor-attached rather than trailing implements.”

My biggest frustration in attempting to apply the Yeomans plan to this farm was the Keyline pattern cultivation itself. As mentioned above, Yeomans himself realized that it didn’t work in every primary valley. Whether it did or did not work, what Yeomans failed to mention was that one of the benefits of Keyline design—making regular, machine-friendly patterns on the ground—actually backfires in a complex landscape. If each primary valley gets its own cultivation pattern derived from its own Keyline (some of which won’t work and will require a herringbone pattern), and if each primary ridge gets its own cultivation pattern derived from its own unique contour reference line, then this farm would have no less than eight separate ridge cultivation patterns. It would have seven or more valley cultivation patterns depending on whether the Keyline geometry actually worked in that particular valley, whether you classified one “sort-of-kind-of-possibly-a primary valley” as one primary valley or two primary valleys, and whether the herring-bone pattern needed to be applied.

About the Author:

Mark Shepard heads Forest Agriculture Enterprises and runs New Forest Farm, an 106-acre commercial-scale perennial agricultural ecosystem that was converted from a row-crop grain farm. Trained in mechanical engineering and ecology, Mark has combined these two passions to develop equipment and processes for the cultivation, harvesting and processing of forest-derived agricultural products for human foods and biofuel production. Mark is a certified permaculture designer and teaches agroforestry and permaculture around the world. 

Also by Mark Shepard:

Titles of Similar Interest: