Testing a Different(iation) Approach to Permaculture Design Process – Part Two: An Example

Carrying on from the previous post, here we share an experimental example of permaculture design when conducted explicitly as a process of differentiating a pre-existing whole into parts.

In November 2015, we collaborated with a team of eighteen co-designers to complete a design process with two clients (Mitch & Mel) and their 55-acre small farm holding north west of Melbourne, Australia. Our 18 co-designers were participants on a two-week intensive permaculture design course where completing a real design project was a requirement of the course.

Though we’ll summarise key aspects of the process as a whole, we’ll focus primarily on how we used a differentiating process to gradually unfold and refine a landscape layout particular to the context of these people and this place.

Note that this example is shared as nothing more than one early experiment in approaching design as a differentiating process.

The People

Aside from establishing rapport and a healthy working relationship with the clients, our process initially focused on clarifying a destination for the project along with identifying, filtering, condensing, and sequencing the key desired areas the design had to accommodate.

Clarifying a Destination

Due to their availability, we met with the two clients separately on successive days.

Franklinford Interview Mel

The team interviewing Mel on day one. No shortage of note-takers.

A bonus was focusing on what they each individually wanted out of the property first and only then accommodating these two (overlapping but distinct) sets of desires into the articulation of an overall destination for the project.1

We gradually transformed a relatively raw collection of desires:


Into a relatively refined draft articulation of what Mel and Mitch wanted to be true of their place:

Our place honours a diversity of natural landscapes thriving with wild areas, flowing water and productive, manageable fields and forests. We are abundantly self-sufficient as we work with pride to nourish and rejuvenate our land into a sanctuary that cocoons us and all its inhabitants in life and serenity

This statement was not what we wanted for Mel and Mitch. It was our attempt to hone in on the essence of what they wanted for their lives on this property. After receiving their enthusiastic approval, we now had a provisional destination for the project – something we could use to guide the rest of the design process toward.2

Desired Areas


Based on what we’d all picked up from our conversations with Mel and Mitch, we next listed a bunch of different areas or activities they wanted to incorporate into the whole of their life on the property.

Rendered Franklinford Wishlist

From the perspective of trialling a differentiation-based approach to design, this was a dangerous moment. Why? Because we appear to have a bunch of elements just begging to be assembled! As explained in an earlier post:

…until we appreciate the distinction between a generic potential part and an actual designed-in part, it is easy to mistakenly think of the elements on a wish list as already defined actual things or elements. From there it is hard to avoid the trap of construing the design task as assembling these prematurely defined elements. The process of design is then inevitably understood as the process of combining these elements where, if anything, the parts birth the whole.

We circumvented this trap by treating these things not as pre-existing parts to assemble into a whole, but as desired areas and activities to be identified or differentiated from within the already-existing whole. Only after this happened would they move from being generic potential parts to actual parts of the system being designed. If you like, you could say that these items are nominees or applicants for the position of being part of the target system. Part of our role as designers is to interview them and find out if, and if so in what form, they might be suited to the job.

Trap avoided, we still had the problem that an individual act of differentiation starts with one thing and differentiates it into two things. Yet looking at this collection of almost 30 discrete items, we can easily become overwhelmed and confused about where to start. Further, many of the things lure us into jumping right into a level of detail or resolution that is premature. The trap we were attempting to avoid again invited us to pull out the base map, grab something, pop it in somewhere, grab the next thing, place it relation to the first thing, and so on. But again, in the interests of trying something different, we resisted.

Instead, before moving on, we refined the wishlist into a form conducive to a differentiating process (as opposed to an assembling process).

Refining the Wishlist (via Filtering, Condensing & Sequencing)

In this step we moved from nearly 30 things to just five high-level areas. Along the way some things got filtered out, somethings got condensed into broader areas, and then these broader areas were sequenced with reference to a simplified version of Yeomans’ scale of permanence:

  • 1. whole farm water system (harvest, storage, distribution including water features)
  • 2. access ways for animals (humans and other livestock) and vehicles (farm bikes right up to heavy trucks)
  • 3. trees (further clustered into a. everything but nuts and b. nuts)
  • 4. the homestead (the details of which, such as the house, orchards, parklandish gardens, domestic animal housing etc etc we put aside for a subsequent and more specific design process for that area once the mainframe pattern for farm as a whole had revealed itself to us)
  • 5(a&b). pasture or open land generally including an area for the future market garden

Here’s the diagram we used during this process…

Wishlist Cluster

…where the understanding was not that we were assembling primary elements into secondary sets, but that we were revealing the deeper patterns or areas the initially listed items already naturally sat within. To compensate for the silent monopoly of element-assembly thinking in our (perma)culture, this point deserves emphasis.

So, for example, with the trees our thinking wasn’t so much “oh – great, we can assemble the shelter belt and the woodlot and the nut groves and form one larger tree system with them.” It was more “okay so these would all be aspects sitting within the whole-farm tree system – if we figure out roughly where that whole system is going, we can worry about chopping3 it up into woodlot, shelter, etc, later on. But let’s get the overall pattern of the tree system as a whole right first.” A subtle distinction, perhaps. A subtle distinction with significant design process implications.

Another point is that though we sequenced the main areas according to the scale of permanence, in different design process contexts different sequences will be appropriate. The main thing is to carefully design the sequence such that you can make earlier design decisions/differentiations that will not be upset by later decisions/differentiations and thus require you to backtrack and completely undo something you’ve already done. It is also important to not get attached to whatever sequence you create, such that you can adapt the actual sequence itself as appropriate as the process moves forward.4

Something Begins to Stir

In a real sense, even though we hadn’t yet tuned into the actual physical space in any detail, we were getting, simply from the destination and the identified, filtered, condensed and sequenced desired areas/activities, an inkling of what the whole might some day include and feel like. We could start imagining what it might be like to be here, based on the sorts of areas involved and the desired overall feeling for the place. More so than when we started, we were already able to garner a fuzzily defined inkling of what was to come.

The Place

Having made a solid start on tuning into the people, we moved our attention to the place.5

During this stage we got a lot done in a short time by breaking up into a series of small teams, each with a specific landscape-mapping task.


Here is the site aerial photo and boundary (dotted-low resolution contour lines can also be seen, bottom-right you can see the current building envelope in white). Folk from the northern hemisphere please note that north is up in base maps from this part of the world.



One team mapped the directionality and seasonality of incoming energies such as fire, wind, water, & views:6



One team mapped the place according to the overall topographical features. Here is a low-resolution summary:



One group did a lot of digging and mapped the boundaries between the two main underlying geologies on the place (note the close but imperfect correspondence of the main ridge and the basalt section):



Though we’ve lost track of the map, another group mapped the relative depth and type of topsoil right across the place.


Another team mapped water movements:


Land Units

Below is a very rough sketch we drew to get across the sort of thing we were after here – in a sense the culmination of tuning into the site – which is combining all the previously garnered information along with carefully walking the site to map the existing pattern of differences across the site. Sometimes something like this is called area analysis, or land unit, component, or system analysis.7


Essentially we are differentiating the landscape based on its own existing character.8 The more we can be with the place with an open, listening, inquiring spirit, the more of this character the place reveals to us. David Holmgren (1994) nicely explained both the nature and point of this process:

Identifying underlying natural land types reveals different qualities. For example, what looks like just a field may have well drained and poorly drained areas. These differences form a basis for land division into basic units.

The boundaries between land types then become lines for infrastructure such as access, shelter belt etc. It also makes it much harder to do stupid things, (like a tractor bogging in a wet corner).

Landscape laid out according to natural land forms has an underlying sense to it that makes it more likely that people will manage it sensibly.

Broadacre landscapes that have lost hedgerows offer incredible opportunities not only to put those features back in but to put them back in a different place, that is, according to the underlying structure, soil types etc. With infrastructure like shelter, access, water etc. we should try to reinforce those natural boundaries; when we do, we create a landscape that has a natural harmony.

When you only have one product or crop, the logic is to try to turn all that land into a place for that crop. When you have multiple land uses like forestry, aquaculture, grazing, horticulture, you can find a place for everything in that landscape.

The team responsible for mapping land units first went through coarsely:


And then in more detail (there was a detailed note for each number on a separate page detailing the unique quality of each spot):


Unfolding the Design

We were now ready to move more explicitly from what was to what might be. We now had:

  • a destination statement for the whole (the property including the people)
  • a list of high-level desired areas and a base sequence or order for dealing with them
  • a satisfactory discernment of the existing pattern of differences or land units across the site

Take One

The moment of truth had arrived. We handed out copies of the base map to everyone in the design team. We reviewed all that had come before. We each picked up a pen. We each separately sketched in a provisional configuration for the mainframe water, access, tree, pasture and homestead areas.

In doing so, we were consciously differentiating the whole space into sub-spaces.

Bringing our various sketches together, many of us were surprised to find something like a 90% correlation in the overall pattern of what we’d each drawn. Here is the rough sketch we then made to summarise what the various sketches tended to agree on:


Let’s walk through the process of of generating this rudimentary first pass.9 As vague as it looks, as a few messy squiggles, herein lies the rudimentary kernel from which all subsequent steps in the design process unfolded.

First Stroke

Starting with water 10 and access considerations (for reasons discussed earlier), we first drew a line representing a proposed layout for the requested new driveway:11


This layout corresponded (more or less) to both:

  • the break of slope between the main ridge and the surrounding flatter land
  • the change in soil that roughly ran along this same line

This meant that:

  • it wasn’t too steep to make construction overly difficult/expensive or too low so as to get too wet in winter
  • it allowed the entire property to be seen upon entering or leaving (making it easy to keep an eye on trees and animals etc)

The upshot was that we’d simply drawn a line highlighting the already-existing and prominent line of  difference between two primary land units, areas, or parts of the site:12


Second Stroke

We subsequently moved our attention to what we’ve here called Area 2. We in turn differentiated this into an area lending itself to trees (Area 3), and areas that lent themselves to more open pasture (Area 2). The primary line of delineation between these two new areas was, again, a pre-existing line, edge, or boundary on the site; in this case the spring-fed creek running through the top of the property in a west to south-westerly direction.

As with the driveway, there were sound functional reasons for this delineation. One was that, once established, the tree systems need less (visual and physical) attention than the livestock systems, thus making sense to be further away. Another was that in this location the treed area provides desirable shelter to the pasture area from the prevailing hot summer winds and hence a primary direction of fire risk.

Third Stroke

About here it occurred to us to differentiate an extension to Area 3 (trees) that would pick up and include the riparian strip running along the creek marking the south-western boundary of the property, while extending right around to provide a shelter belt and wildlife corridor around the south-easter perimeter. We’ll call this Area 3b.

In this case, unlike the examples above, the line of differentiation here was not about accentuating a naturally pre-existing demarkation in the landscape. Here it was more about running roughly parallel to, and very roughly about 15-20m out, from the property boundary. In this sense, its form was defined by a more arbitrary pre-existing feature – where surveyors had previously happened to have located the property boundary. For this reason we indicate it here with a slightly fainter and dashed line (that would likely become a fence in future).

Once again, this line of differentiation made functional sense from a shelter and wildlife corridor perspective, and spoke especially to the client’s desire for the place to feel like a cocooning sanctuary.

Fourth Stroke

About here we focused in the now slightly smaller Area 2, which we now differentiated into two additional areas – one that lent itself to more extensive pasture (Area 2), and one that lent itself to more intensive uses such as horticulture, market gardening, and possibly the planned homestead (Area 4).

In this case the line of differentiation was also more subtle, coming down both to a possible place for a vehicle access path to the northern part of the property (and a pre-existing gate there) and also the extent of the more naturally sheltered, richly soiled pocket of land to the east of this line.


To summarise, we had drawn four lines across or within the property. In doing so we had differentiated four high-level areas (five if you were to treat 3a and 3b as distinct).

Note that the lines of differentiation…

…and the areas differentiated…

…are like the two sides of the same coin. You can’t have two areas without something differentiating the one from the other, and you can’t have a differentiation without some areas being differentiated!13

The Plot (not to mention the driveway) Thickens

Let’s now come back to the driveway, the creek internal to the property, the second-tier vehicle access way or track to the northern part of the property.

The plot thickens when we acknowledge that these lines, edges, or boundaries not only distinguish different areas from one another. They themselves constitute areas (if long and relatively thin ones) in their own right.

In other words, we had not in fact only differentiated four big blob-like areas. We had differentiated four big blob-like areas and three narrow stretched-out areas. So after drawing just three lines we had in fact differentiated seven discrete areas!14

A Last Few (More Detailed) Differentiations within the Driveway Area

Now take the driveway as its own area. During the process we zoomed our focus into this area, and, as indicated roughly on the raw first concept design shown above, differentiated it into several subsidiary sub-areas.

Firstly, as much as a driveway it constituted a water catchment surface and diversion drain that would allow runoff from the main road outside the property as well as the new driveway itself to be harvested and gently directed to a storage point high in the landscape (which we didn’t get as far as highlighting on in this first pass).  As explained in a previous post, Darren J. Doherty calls this particular patterning of driveway and drain an in sloped gradient catchment road:


So as shown in the top of the diagram above, here we differentiated the driveway area in two sub-areas: the road itself and the drain next to (and uphill of) it.

Secondly, along with the adjacent row of deciduous, fire-retardant trees it made sense to plant directly below (and possibly above) the new driveway, the driveway (as an un-vegetated and hence non-flammable) surface made a useful contribution to resilience to wild-fire (a priority in this area).

So now, the driveway, while initially differentiating the property as a whole into two broad areas, itself constituted an area containing several subsidiary areas – those taken up by the fences, trees, drivable surface itself, and drain (here just showing the option of just a single fence and row of trees below the driveway):

Taking this into account, if we include the road, drain, and ribbon of trees as sub areas within a (single or double) fenced strip running around the ridge, and if we include the internal stream and secondary proposed vehicle access track as areas in their own right, we now have 7 high-level areas some of which contain internally differentiated subsidiary areas (3 & 5):

All in all, using a sequential differentiating process, we had generated a tentative preliminary configuration of areas – a high-level pattern for the design. Our next task was finding out what was wrong with this configuration by getting out there and crash-testing the map against the territory.

Take Four

In conjunction with revisiting the site and bumping our vague preliminary initial ideas up against the reality of the site, we went through a very similar process three more times. Each time the overall configuration or form of the design got better. Each time the parts became more refined, organ-like, or organised. They hung together with more and more coherence on each subsequent pass.

The key focus was walking and scanning the site for conflicts, tensions or issues that whatever version of the design we were up to either had failed to resolve or had actually created. We were effectively searching for places where the reality of the site trumped or dictated updates in our designed ideas about it. Whatever we did next in the process then attempted to resolve those tensions.15

Skipping versions two and three, here is the fourth version we got to after much conversation, re-walking, and re-checking.

We won’t bore you by repeating the entire sequence of differentiations generating this more refined configuration of parts. In essence it was the same process, but by benefiting from what had come before it resulted in a layout better adapted to both the site and what the clients wanted their life there to feel like:

Our place honours a diversity of natural landscapes thriving with wild areas, flowing water and productive, manageable fields and forests. We are abundantly self-sufficient as we work with pride to nourish and rejuvenate our land into a sanctuary that cocoons us and all its inhabitants in life and serenity

The clients felt the potential of the design to help them realise their vision, and loved what we’d come up with. In their words it positively altered their whole perspective on their property. As far as implementation goes, in the last few weeks the new driveway has been fully installed and Mitch and Mel are delighted with the outcome (A likely outcome in our experience when legendary local permaculture earth mover Graeme Jennings is behind the wheel):

This is not to say what we got to is perfect. It is not. But as far as the design team and the clients were concerned, it was a solid step in the right direction, and sufficient to guide their next steps in developing the place, during which they will get feedback allowing them to further refine what happens next.

David Holmgren (2002) writes that:

Complex systems tend to evolve from simple ones that work, so finding the appropriate pattern for that design is more important than understanding all the details of the elements in the system

We hope that with this example we have unfolded something in the ballpark of an appropriate pattern for this site.

We’ll add an aerial photo of the new driveway when it becomes available, in which no doubt we’ll see further improvements on where we got to with the design that came out of the process of adaptively implementing that part of it.


We have just seen in some detail an example of an approach to permaculture design flowing from the alternative approach to permaculture design we have been exploring in previous posts (starting here). It is true of this example that it:

  • Started with an existing configuration of a whole-space-comprising-a-configuration-of-already-differentiated-parts…
  • …further differentiated this whole…
  • …fluidly moving down, up, and sideways as necessary…
  • …both modifying what is there and conceiving (as potential) then introducing (as actual) new parts…
  • …that grow out of and hence harmonise with the whole…
  • …to support the evolution of that whole…
  • …as a rich network of interrelated parts…
  • …toward our desired outcomes of a resilient, abundant, human-supporting ecosystem (and specifically Mel and Mitch’s vision statement)

Diagrammatically the process was clearly more an instance of this:


than an instance of this:


To close, recall that part of the prompt for sharing this example this comment from renowned permaculture author Toby Hemenway:

Now, if only someone would develop a methodology that shows how one can do what Alexander is suggesting…

For that matter, a different version of the same prompt had earlier come from permaculture co-originator David Holmgren. While not dismissing the utility of integrative or parts-towards-wholes thinking and design, David agreed that, in his words (bold italics added for emphasis):

  1. there is a huge cultural bias towards details to pattern understanding and designing
  2. nature works from pattern to details
  3. we need most effort into creating design processes that effectively achieve this second pathway

Well, Toby, David, hopefully you agree that this is one broad-strokes example, however rudimentary or flawed its details may be.

Being that one example is never enough, however, in our next post we’ll go ahead and share another.


Holmgren, David. “Ethics and Principles of Permaculture” (transcribed from a workshop recording by Chris Dixon). August 1994.
Holmgren, David. Permaculture: Patterns and Pathways Beyond Sustainability. Melliodora, 2002.


  1. We thank Dave Jacke from whom we’ve adopted talk of articulating rather than stating goals.
  2. Some readers might detect a flavour of holistic decision making here
  3. excuse the pun
  4. As we saw in the example of Darren J. Doherty’s design process, the reality is that you jump around a lot more than follow the prescribed sequence precisely. It is a fluid, malleable template.
  5. Note that with each additional layer of inquiry into the place as a whole, we not only tune more deeply into what is, but, as we’ll see later, we simultaneously gain small increments in clarity about the shape of what might become. As any experienced permaculture designer will insist, a good design must arise up from, honour, and hence reflect the (sometimes subtle or latent) patterns pre-existing in the landscape. These patterns are pre-existing differentiations in the landscape, and are what define its unique character.
  6. Keep in mind this was their first sector map if you notice some slight inaccuracies in the summer and winter sun sectors
  7. See for example David Holmgren’s discussion of the land systems approach in this PDF essay
  8. Technically, we are differentiating differentiations
  9. Keeping in mind with the description that follows that the fluid, highly dynamic, and multi-streamed reality of how these patterns emerge defies any simple linear, logical sequence
  10. We’ll explain how this was true later
  11. We draw this line as an uni-directional arrow here to signify the act of drawing it – as an actual driveway, however, traffic would flow either way
  12. Compare with David Holmgren’s statement shared earlier that “The boundaries between land types then become lines for infrastructure such as access…”
  13. Another way of saying this is that every thing has an edge.
  14. Note that here we will leave out the dashed line demarcating the suggested perimeter tree area on the southern half of the property. The reason is that though it makes sense to plan a fence here, which is a (very thin) area of types, this thought did not cross our mind at the time, whereas the proposed access roads and the stream as distinct areas did. In the case of the perimeter tree area the line of distinction was more simply a conceptual tool, like the equator
  15. A simple example was how the shape of the proposed new driveway changed from version one to four so as to accentuate, rather than slice in two, the area best suited to the homestead and market gardens


  1. I am curious about your thoughts here – when dividing the wholes into parts to see what is conducive for what part (element) there is sometimes a not so clear answer.

    An example might be in deconstructing a whole site by soils we see one area is clearly better than another in terms of supporting growing vegetables. Lets say this ideal garden area is in what might be zone 4 or 5 in relation to the primary residence. Granted we can locate a garden in zone 1 or 2 and build soil over the course of years but that seems to not necessarily outweigh the energy of a daily venture to zone 4 or 5 to tend a garden in soils that need minimal development.

    How do you reconcile conflicts in design such as this? When the whole broken down into parts doesn’t align with other permaculture principles?

    1. Hey Bret. Awesome question. The thought that comes up for me is that at a high level I see the point as tuning into people and place deeply enough to fathom and facilitate the emergence of a configuration that honours and reconciles the needs/inclinations of both. As you say sometimes a conflict crops up where it seems like we must choose between honouring or prioritizing one vs the other. I personally dislike the attitude of “oh well, let’s find the best compromise” and instead strive to go deeper in seeking a configuration/solution/direction that genuinely accommodates both. I’m not saying that in some cases compromise isn’t required, but in my experience such ‘hitches’ often turn out to be opportunities prompting solutions or developments where people and place both end up better off. Does that make sense? As for dividing wholes into parts I think that while an overall movement from patterns (wholes) towards details (parts) is desirable if we’re to mimic nature, that the reality of any healthy design process will involve division and subtraction as well as addition and multiplication as sub-instances of gradually transforming a people-site ensemble in desired directions (see this comment for more on how my thinking has evolved here).

  2. This seems very much like what I have called “finding the rooms in an outdoor space”. The natural tesselations of the patterns on the landscape. Mollison hints at this in his Designers Manual chapter on patterns but never takes it any further.

    1. Thanks for your comment and that’s a lovely way of putting it Milton. I must track Mollison’s hint down – ah yes – page 74.

Leave a Comment

Your email address will not be published. Required fields are marked *