Published by City Farmer, Canada's Office of Urban Agriculture


Urban Agriculture:
The Potential of Rooftop Gardening

Chapter 3: Observations

By Joseph St. Lawrence
Joseph St.Lawrence

Report of a Major Project submitted to the Faculty of Environmental Studies in partial fulfilment of the requirements for the degree of Master in Environmental Studies.

York University, North York, Ontario, Canada.
July 29, 1996 [Images are not included here.]

Chapter 2 (The Site and the Plan)
Chapter 3 (Observations)
Chapter 4 (Conclusion), and References


Even the best laid plans can fail to produce their expected results. This is particularly true of agriculture, which is so dependent on external factors. One learns from ones mistakes however, and trial and error is the only way to integrate an agricultural enterprise into a local environment. I am pleased with my successes on the rooftop, but there are a few things I would do differently in the future. My observations and thoughts concerning the growing season are recorded in this chapter.

Cropping Plan

The cropping plan needed some modification during the course of the project. This was to be expected, given my own inexperience with both the site and some of the crops. Unfortunately, change was not always easy, as I was more-or-less locked into a four week cycle, and altering the contents of the garden beds was impractical during the growing season.

My weekly planting schedule ran as follows. On Monday, a new set of transplants was planted. Soil blocks seeded the previous Wednesday were moved to the watering table against the windows, and the blocks already on this table were moved to flats to begin their hardening-off procedure. Tuesday was used for general maintenance and making blocking mix if necessary. On Wednesday, a new set of blocks was made and sown on the seed starting table. Thursday was another general maintenance day. On Friday, greens were harvested and washed. If rain was likely the following Monday, transplants were also planted. If not, the seedlings were left outside for the weekend to finish hardening-off. If it did not rain on Friday or during the weekend, Saturday and Sunday trips to the warehouse were necessary to ensure that the hardening seedlings did not dry-out. (The soil blocks are fairly good at retaining moisture, and can go a day without water, but I took the precaution of checking on them regardless.) The total amount of time required for this work was between ten and twelve hours per week.


I was pleased to observe that most of the crops chosen did quite well in the 3" of soil they were provided (a more detailed discussion is found under Crops, below). As crops were removed, nutrients were replenished by the addition of a few handfuls of blood meal and kelp meal. There were no obvious nutrient deficiencies manifested in the appearance of the plants.

Crusting was a small problem, and was confined to those beds lacking mulch. It was problematic because the hard surface prevented the quick absorption of water and encouraged run-off. It was easy to remedy this with cultivation when the plants were widely spaced, but where seeds had been broadcast, this proved difficult.

Organic matter in the soil gradually increased when crops were harvested and roots were left behind. It is unfortunate however that there were no worms to take advantage of this (although sow bugs, another decomposer, became quite plentiful). I considered inoculating the beds with a few red worms (the type of worm used in composting) but decided against it. I was worried that they would cause problems if the soil were moved indoors in the fall, or they would freeze to death in the 3" beds if left outside. In any event, the soil texture did improve gradually over the course of the season (both from the organic matter, and the peat moss supplied by the soil blocks).

Beds and containers

The garden beds held up quite well, and displayed no obvious signs of decay or damage. I had considered using Jeavons's (1991) five foot width for the beds, but was glad I went with four feet, as I was just barely able to access the centre of each bed for planting and cultivating.

One factor I neglected to consider when planning to fill the garden beds was spring rain. The compost on the roof was quite difficult to move and fairly uncooperative in general when wet. I had to work very rapidly during the few dry, sunny periods in order to get the beds filled in time for planting. This was not a great problem, but could have considerably delayed operations if I had planned an extensive spring planting.

The soil barrier in each bed was compromised after the first heavy rain. I thought at first that the compost had simply settled, but some digging revealed that the newspaper had disintegrated. I would therefore advise that another medium be used for this purpose. I experienced no drainage problems, in spite of this occurrence, so perhaps the soil barrier is unnecessary for this type of container.

Shade and mulch

June, 1996 was an abnormally wet and cool month, so I had little need of either shade or mulch. The shading crates did prove useful for protecting recent transplants (especially claytonia -- see below). I found that they were surprisingly good at retaining moisture in the beds, and used them on occasion to protect broadcast seeds from desiccation on sunny days.

I had no idea if, or when, the weather would warm, so I applied straw mulch to the beds as a precaution. The mulch was very effective at retaining moisture. Mulched beds would take three or four days to dry out, whereas unmulched sections would be dry in one or two days. The mulch also had a beneficial effect on soil texture: it prevented the crusting common to unmulched areas. Finally, the mulch kept the soil beneath it noticeably cooler than exposed soil.

There was one problem with the mulch, and that was weeds. It may have been that weeds simply grew better in the cool, moist environment it provided, but I think a significant number of weed seeds came in with it. Lamb's quarters and grasses were the main invaders. The lamb's quarters were fairly easy to keep ahead of, but the grass was always difficult to completely remove.


Most of the crops grew well in the 3" beds. The cool, wet weather during the project summer was beneficial to the growth of many of the salad greens. Outlined below are some observations on the suitability of cultivars, and cultural considerations for a few special crops.


There was a marked difference in the yields from different types of lettuce. The red lettuce varieties were often twice the size of the green varieties found in the same bed. I found the varieties Lollo Biondo, and Curly Oakleaf particularly disappointing, and Red Sails, Red Salad Bowl, Valeria, and Rossa d'Amerique to be very fast growers. I speculate that the red pigments helped these greens cope better with the low light levels during the cloudy season.

Some of the lettuce varieties are very attractive. Rossa d'Amerique, Rossa di Trento, and Valeria are beautiful varieties, and look very good in the salad mix. Their quick growth, appearance, and utility as cut-and-come-again greens makes them ideal candidates for mesclun cultivation.


All varieties of mustard (tatsoi, Osaka Purple, and mizuna) bolted the week of June 17th, regardless of age. The last early season crop of these was planted May 17th, and was harvested before it bolted. Fortunately, mustard flowers and seed pods are edible and add an interesting effect to the salad mix, so some of the bolted mustards were left in the beds to provide these.

Mizuna and tatsoi were the most interesting components of the salad mix, judging by the comments I received. Mizuna has large, thin, and deeply toothed leaves which look very attractive in the mix. Tatsoi has a thick, spoon shaped leaf with prominent white stem. Both of these varieties have a mild flavour, and can therefore be significant contributions to the mix. Osaka Purple mustard, by contrast, is quite pungent and must be used sparingly. Its leaves also tend to get quite large if it is not harvested at an early stage.

All of the mustards grew very well in the cool spring. Much better than the lettuce, in fact. They appear to be good greens for early spring and late fall mesclun mixes.

Other greens

Figure 3.1: New Zealand spinach seeds.

New Zealand spinach and malabar spinach were very disappointing. I was depending on these crops for hot-season greens (they are alternatives to the mustards which will not grow well at that time). They are both difficult to germinate and slow growing. It took a number of months before I finally discovered how to germinate the seeds of New Zealand spinach. The seeds are large, and I read that they require scarification and soaking before planting. This I did, and I put them in plastic bags to pre-germinate. After a month no sprouts had emerged, so I started another batch. Another month passed with still no results. After dissecting some seeds, I found that they are actually berries, with the seeds confined to the flat end (Figure 3.1). I cut the tops off of a number of these berries, and they germinated fairly quickly after planting in soil blocks. Unfortunately, I had lost two months by this time, and the crop did not develop in time to meet its intended need.

Claytonia, also called miner's lettuce, is a very attractive addition to the salad mix. It requires more coddling than the other greens, however, and my ignorance of this caused some early losses. It seemed very sensitive to increased exposure to light, and suffered very bad leaf-burn if hardened-off for only one week with the lettuces and mustards. I found that the plants did much better if partially shaded for a week or two after transplanting. They suffered less burn, and established themselves and matured more quickly than those exposed to full sun from the start.

Mache and dandelion both suffered from slow germination. They came up about a week behind lettuce and mustard planted at the same time, and were too small to transplant when the other greens were ready. This could be remedied by seeding mache and dandelion a week earlier than the other greens, or by adopting the broadcast-sowing discussed below. I will not bother with mache in the future, except perhaps in the early spring. Dandelions, however deserve more attention because they are able to grow in the hot season, unlike most of the other salad-mix greens. I think there may have been a problem with my seed stock, as germination rates for the dandelion were often below 50%.

Arugula and cress grew very well in the early season but suffered as the temperature rose. Cress was particularly disappointing, and the later crops failed entirely. Arugula was somewhat more dependable, and seemed to benefit from partial shading during the hotter days.

Purslane is a very successful hot-season green. It tolerated both the heat and the shallow soil. I found it reached maturity faster than the seven weeks cited in Forsyth & Mohr (1992), and recommend its cultivation for mesclun. I also found that it stored quite well, again in contrast to Forsyth & Mohr. Its succulent leaves are quite delicate, however, and must be washed and dried separately from the other greens.

Finally, we come to the star failure of this experiment, radicchio. Most of the crop bolted by the end of June, and the rest followed by mid-July. This is not likely due to heat or water stress, as June was fairly cool this year, and the crop was well mulched. I think, rather, that the crop had trouble with the 3" soil depth, and the plants sent up seed stalks when their roots became restricted. More experimentation would certainly be required, but I do not recommend this crop for shallow bed cultivation.

Other crops

Flowers proved to be a very inefficient use of space, being slow to mature and requiring wide spacing. They should have been started indoors, rather than direct seeded, but space limitations prevented this. I would recommend that others with similar space restrictions confine their efforts to one or two varieties that do not require much space and that serve other purposes. Arugula and mustard greens, for example, produce prolific edible flowers if left to run to seed. They also produce useful greens prior to bolting. Shungiku, borage, and nasturtiums also produce useful young greens and edible flowers. This combination of greens and flowers will produce a better use of space than the cultivation of crops like calendula and agastache that do not yield edible greens.

No problems were observed with the cultivation of green beans in the 3" beds. I chose a variety of filet beans, which are meant to be harvested when quite small. Unfortunately, the small space allotted to them was far below that needed to produce a worthwhile yield, and I do not recommend their cultivation in gardens where space is precious.

Kohlrabi was another successful crop, although it, like the green beans, requires too much space to produce a decent return in a rooftop garden. It is usually planted at a 4" spacing, but since this would have been very difficult to manage with transplants, I took Coleman's (1989) advice and tried starting it in a multi-plant block. A multi-plant block is a normal soil block which has been planted with more than one seed. In the case of onions, Coleman puts four seeds in each block, and sets-out the blocks at 1' spacing. This gives a final spacing of four onions per square foot, the same as would result from planting individual onions 4" apart in rows 1' apart. As the onions grow, they push their neighbours away and produce normal bulbs. I reasoned that kohlrabi should do the same thing, and indeed it grew quite well using this method.

Okra, cucumbers, and tomatoes did not do very well in the shallow beds (the tomatoes in buckets were fine). The plants seemed healthy, but they were stunted and slow growing. I do not recommend the cultivation of okra where space is at a premium, and would confine plantings of cucumbers and tomatoes to deep containers.


The late spring delayed outdoor planting until May 3rd (I had expected to put my first transplants in the ground on April 15th). Cold weather was only part of the problem, as the soil was often too wet for transplanting. (Indeed, the problem of transplanting in wet soil plagued me throughout June as it rained for 20 of the first 24 days of the month.) I also lost the fallow week that had been a part of the cropping plan as I waited for the greens in the first pair of beds to reach a size large enough for harvesting.

The first harvest of greens was on June 3. Approximately 1/3 of a bed-pair was harvested, yielding 316g of salad mix. On June 6, a pair of beds was harvested, and this yielded 620g of mix. These yields were far below what I had expected. According to Jeavons (1991), leaf lettuces should yield between 0.4 and 1.7 lb. per plant. With 54 heads per lettuce bed, I expected at least 20 lb. (approximately 10 kg) of mix per week. The plants were clearly not reaching expected size in the seven weeks they had been allotted for growth. This may be attributed primarily to the cold, cloudy spring, which undoubtedly had an effect on plant maturity (later greens were somewhat larger). Also, the salad mix contains only leaves, and will consequently weigh less than lettuces, which include stems and part of the root. In any event, these early harvests suggested that a radical change was need to the cropping plan, if any acceptable yields were to be obtained.

Table 3.1 Garden yields
Date Yield
June 6 620 g
June 17 2.2 kg
June 21 940 g
July 1 1.5 kg
July 5 2.7 kg
July 16 1.8 kg
July 18 3 kg
July 24 1.8 kg

The first change involved planting distances (Photos 3.1 and 3.2). The lettuce plants were initially set-out at 8" spacing, and there was a lot of empty space in the bed come harvest time. I decided that I could double my yields by doubling the planting density. I tried a number of routes to this end. The first option was to produce twice as many transplants each week, and plant them at 4" spacing. This was unworkable given the limited space I had for starting transplants, and because transplanting blocks closer than 6" is difficult. A second option was to harvest a lettuce crop as cut-and-come-again rather than digging up the whole bed and re-planting. To increase yields, I simply planted new seedlings between the harvested rows with the old stumps in place (Photo 3.4). I could then get a second and perhaps third cutting from the first set of lettuce plants and also get cuttings from the new set. This proved to be a very satisfactory arrangement, and increased yields once it was adopted (results appeared on June 17th: q.v. Table 3.1). The resulting spacing was 6" between each plant. This eliminated unused space in the beds while still providing each plant with enough room to grow.

Yields could also have been increased by broadcast sowing the beds, and I seriously entertained this prospect for some time. It was unworkable with the cropping cycle I had planned, however. Broadcasting would save the labour and materials associated with starting transplants and hardening-off, and would eliminate wasted space by producing denser stands of greens. Yields could be higher because the lettuces and greens formerly grown in two beds could be grown in one. A pair of beds could still be harvested and planted each week, effectively doubling the yield possible with my initial plan. This, at least, was what I thought until I tried it. Perhaps I lacked the necessary skill at broadcasting mixed seeds, but I found the stand of plants produced to be very uneven and full of gaps. Clearly, this method requires both practice, and much more seed than I anticipated to result in a full stand.

There are a few other disadvantages of broadcasting mixed seed into the shallow beds. The first is that care must be taken to ensure that the seed-bed remains moist throughout the period required for germination and seedling establishment -- this may require two waterings a day on dry, windy days. The second problem is that each pair of beds would require seven weeks to reach maturity rather than four. (Using fourteen beds for the rotation would solve this, and could be done given the current number of planters if fewer planters were devoted to flowers and vegetables.) A third problem is that applying mulch and weeding would be difficult. Mulch would be unnecessary once the soil surface was covered -- indeed, the leaves will form their own "living mulch." Potential weed problems was one of the reasons I chose to avoid this method at the start. The cultivator must be familiar with the greens and local weeds, otherwise weeding could become a slow and difficult task. My brief experience tells me that broadcasting is not the best method of sowing mesclun. The difficulties outlined above, and the fact that one cannot easily harvest a single crop if necessary (arugula or lettuce, for example) make separate plantings the most appropriate technique to my mind.

There are a few other changes that were made to the initial cropping plan. Excellent germination rates meant I needed fewer soil blocks than I originally called for: four blocks less were seeded for each plant each week in Table 2.2 (e.g. Block 8 contained 32 Lolla Rossa lettuces, 32 Salad Bowl lettuces, 20 claytonia, and 20 dandelions). I also omitted kohlrabi, mache, and dandelion from later blockings, for the reasons outlined under Crops.

Soil blocks

Soil block making (blocking) is a fairly straight forward task. I used a 500g yoghurt container to scoop the dry blocking mix into a recycling box. For every three scoops of mix, I added one scoop of water, and stirred thoroughly. The block maker was then pressed into the moist mix, any excess was scraped from the bottom, and the blocks were ejected directly onto the capillary watering mats.

I used tweezers to plant the seeds, one per block. It took about 40 minutes to make and seed 120 blocks, working at a leisurely pace. This required about 30 scoops of mix, so each 500 ml scoop yields approximately 4 blocks.

The blocking mix ingredients changed early in the project. I initially used sand for drainage, but soon discovered that carrying a 66 lb. bag of sand on my bicycle was not at all enjoyable. I switched to vermiculite, which is far lighter, and found it to be an acceptable substitute. Coleman (1989) however, warns that vermiculite can be subject to compacting when blocks are being pressed-out. Perhaps this is more of a problem with the large floor model blockers, because I did not notice any problems with my hand-operated unit.

Although I was quite impressed with soil blocks in general, there were a few problems associated with their use. At 2" per side, the blocks require considerably more space per seedling than traditional flats or cell packs. This proved to be a problem at this site because the limited amount of window space for the seedlings. Another difficulty, and this is true for all methods of starting transplants, was the time and attention required for hardening-off the seedlings. They require daily attention during the early phase of this process, and I occasionally found it difficult to keep myself busy while waiting for two or three hours of hardening to pass. It was also somewhat awkward transplanting claytonia, mache, and mustards at the 6" spacing they require. The two inch blocks are simply too large to make this a smooth operation. Finally, I have concerns about the use of peat and vermiculite in the blocking mix. Peat is a non-renewable resource and vermiculite requires a considerable energy investment in its production. It is necessary to find alternatives to these products -- leaf mould and sand are two options -- if blocking is to become a sustainable means of producing transplants.

On the plus side, the blocks did provide a great deal of flexibility with scheduling transplanting. Because there is no root disturbance, the blocks can be planted in the blazing heat of mid-afternoon, with no ill effects. They also conserve garden space outdoors while the plants are starting in the blocks.

I am still of two minds regarding the use of soil blocks for the main greens crop. (I will definitely continue to use them for herbs, flowers, tomatoes, and other transplanted vegetables.) The convenience of broadcasting is attractive (i.e. there is no block making or hardening-off required), but given the problems weeding and mulching broadcast beds and the space restrictions at this site, the use of blocks is, I think, the better method.


Taking Lee's (1993) suggestion, I delayed any intensive marketing efforts until I had an example of my produce to display. After the first harvest, I realised that there was no way I would be able to meet the demands of even a modest restaurant or store. Fortunately, when the director of Field to Table saw the salad mix, she said she would take whatever I could produce, hoping to incorporate it into the Organic Food Box, or use it for catering contracts.

I did have an initial concern about this arrangement. I felt at first that by selling to Field to Table, I was avoiding marketing the product in "the real world." But upon consideration, I realised that Field to Table is as real world as any customer, and has the same demands for quality and price as any other. I was fortunate, however, that their needs are very flexible, and they would not be left short if I was unable to provide a harvest on a given week.

The more I considered the arrangement with Field to Table, the more sense it began to make. Integrating rooftop production and the business beneath the roof seems the only way to take advantage of the benefits rooftop agriculture provides. I often found that I only need to do 10 or 20 minutes of work at the warehouse on some days, but still had to make a 40 minute trip to do it. This wasted time could be avoided easily if the roof space were cultivated by someone already at the site.

A second buyer for the rooftop produce was found in early July. Juice for Life, a local food bar, became a regular customer purchasing lettuce and edible flowers to extend their own salad mix. I received calls from a few other restaurants interested in my produce but I did not have the volume to meet their needs as well. If Field to Table continues to use and expand the site and engages local restaurants in the crop planning, the garden could prove fairly lucrative.


The first two harvests (June 3 and June 6) were used to test different types of packaging and the shelf-life of the product. I chose zip-closure plastic bags, following the lead of other mesclun sellers. I tried two types of bag, one of standard plastic, and a second, designed specifically for vegetables, with small perforations in the plastic to allow for moisture exchange. Plain plastic proved far superior. The greens lasted in fresh form for over a week, whereas those in the perforated bags began deteriorating after three days -- broadleaf cress was the first to go, and turned yellow quite rapidly (it may therefore be a good plant for testing other packaging methods).

Washing and drying the leaves was a time-consuming process, requiring about an hour and a half for a 1.5 kg batch of greens. The greens were washed in three water baths. The entire mix was washed in the first two baths, and large handfuls were washed in the final bath so that any remaining weeds, damaged leaves, straw, insects, and dirt could be easily spotted and removed (most of these contaminants were gone at this stage, but a few blades of grass and straw usually appeared here).

I looked into purchasing a hand-operated industrial-size salad spinner, but the $219 price-tag was prohibitive. Instead I used a mesh bag, salvaged from a shipment of onions (the 50 lb. bags are quite large, and perfect for spin-drying greens). I found that lettuce and arugula leaves were easily damaged by vigorous spinning, so I adopted a procedure of gently spinning a small quantity of greens, and frequently re-distributing them in the bag to speed the drying process.


Flora and Fauna

Weeds were a minor problem, and I believe the mulch I used was the most significant vector of contamination. I was unable to apply the mulch thickly enough to achieve any weed-suppressing action, so it was perhaps more of a problem in this regard than anything. Weeding was by no means onerous, and only took about 10 or 20 minutes every few days if I stayed ahead of it.

I had no idea what to expect in the way of garden fauna. Would I be invaded by pests? Would there be any bees for pollination? Would pigeons eat all my seedlings? These questions weighed heavily on me as the season began.

I am happy to report that there were almost no pest problems at the site. The growing medium was almost free of pests (I found only three cut worms). Most importantly, there was no damage from flea beetles. Brassicas (a family which includes my crops of mustards, arugula, Chinese cabbage, and kohlrabi) are usually plagued by these pests. The mizuna (a mustard) I grew at home was punctured with thousands of holes from these beetles, and my rooftop mizuna would have been unusable if subjected to the same invasion. There are either few flea beetles in the area (unlikely, given the wild overgrown properties nearby), or they were simply unable to find my rooftop greens. In any event, their absence was a blessing.

A number of beneficial insects were found in the garden, including numerous species of spiders, wasps, bees, and flies as well as ladybird beetles and their larvae. These no doubt played an important role in keeping aphids and the larvae of butterflies and moths in check. A few of these pests were spotted, but they did not reach problematic levels of population.


When planning this project, I expected to get a jump on the growing season because of the heat-effect in town, and the heat collecting potential of the large wall to the north of the garden. It was a very cool spring, however, and it is hard to assess the benefits of these factors. The main difficulty lay in the fact that the air was quite cool, and there was significant cloud cover, so there was little opportunity for the walls to absorb much heat and produce any noticeable effects. In fact, it often felt cooler on the roof than it did at ground level because of the increased exposure to wind. Later in the season, I did notice a warming effect from the north wall. Plants and soil against this wall began to wilt and dry out much sooner than those only a few feet away. The heat-gain thus proved to be more of a problem than an advantage in this situation.

I expected to have problems with excessive heat, but these did not materialise. Temperatures in June were consistently below normal however, so in another season this may be more of an issue. At the soil level, mulch was very effective at maintaining cooler temperatures than those found in exposed areas of soil.


Wind was a small problem at the edges of the garden. A few early tomato transplants suffered from this exposure, but after being moved to the centre of the garden area their appearance improved dramatically. Most of the heavy breezes came from the north-east, and I affixed small shelters to the tomato, eggplant, and tomatillo planters to avoid damage to the plants (Photo 3.3). The garden beds were not subject to much wind because of their low profile and the protection of the south-facing wall.


Water proved to be less of a problem than I had anticipated. In fact, the trouble came at the other end of the spectrum, i.e. there was almost too much water. It rained for 20 of the first 24 days of June, and it was often difficult to get transplants into the sodden soil on schedule. As the season began to dry out, mulch took on great importance. Mulched soil, as mentioned above, dried out much more slowly than exposed areas and did not form a surface crust. Mulching would be vital to any similar enterprise if only for the reason that the gardener can take the occasional day off without worrying about desiccation of the crops.

I must state here my concern about the use of municipally supplied water for gardens. As I mentioned earlier, I hoped to harvest rain water for this garden but could not find any practical means of doing so. I estimate that over the course of a 20 week growing season, about 3,200 gallons of water would be required for a project of this size. The cost of this water amounts to less than $13.00 (the Toronto rate is $3.95520 per 1000 gallons), but I feel it is an inappropriate use of potable water. Finding a means of harvesting and storing rain water should be a priority for anyone interested in developing an urban garden.


Early in the project, concerns surfaced about urban air pollution and its potential effect on the safety of the salad mix as a food product. According to David McLaughlin (1996, personal communication), the Phytotoxicology Investigator Coordinator at the Ontario Ministry of Environment and Energy, Standards Development Branch, air pollution is unlikely to have any influence on the safety of the crops at this site.

McLaughlin's branch has been researching the effects of air quality on plants since 1968. The data indicate that most of the toxins collected by plants comes from the soil rather than the air. In the last five or six years, the amount of airborne lead (the primary contaminant of concern) has dropped significantly with the phase-out of lead in gasoline. McLaughlin states that, unless there is a nearby industry or other point-source of emissions (a hospital incinerator, for example) there is no reason to be concerned about contamination of urban garden produce from the air.

Soil is another matter however. As noted earlier, most of the contaminants found in plants are absorbed from the soil. Lead based paint from old playground equipment, houses, and fences, and emissions from the burning of leaded gasoline can all accumulate and remain in the ground. Soil should be tested for contamination before any cultivation is planned. If lead is found, it does not mean that the site cannot be used, but it does suggest that only certain crops should be grown, or that other steps should be taken to avoid the contamination of produce (e.g. replacing the soil, or planting the garden in containers of imported soil or compost as was done on the project rooftop).

Roof loads

Using snow load to support the garden planters poses one big problem: viz. what do you do in the winter? One option is to remove the planters entirely. The second possibility is to displace snow from the garden area using a tarp or some other structure designed to prevent snow accumulations in utilised spaces. Both of these options are being considered. The former is the least attractive, not only because of the work involved but also because a) it will delay the commencement of cultivation next spring, b) it prevents the development of any sort of stable soil ecosystem in the planters, and c) it prevents the use of perennial herbs and flowers in the garden. The same quantity of soil did stay on the roof throughout the previous winter, but I am reluctant to leave it there this year in light of what I have learned about the strength of the roof. The tarping idea may prove difficult to realise, but I hope that the problems can be resolved before November, when a final decision will have to be made.

Social aspects

As mentioned earlier, the management and staff at Field to Table were very supportive of my efforts on the rooftop. Urban food production marries to the organisation's interest in food security. The potential synergy between the rooftop garden and the other initiatives at the warehouse are particularly interesting. All of these features made for a comfortable social environment.

The volunteers involved with the Good Food Box packing and the many visitors to the warehouse often displayed a keen interest in my activities on the rooftop. Unfortunately, however, much of the potential educational benefit of the project was lost because many of the visitors were discouraged from viewing the site because of the ladder climb and the crawl through a window. I did get some media exposure however, with a brief article in the July/August issue of the architectural magazine Azure, a glimpse of the garden in a Field to Table feature on City T.V. on June 17th,, and a feature in the Canadian Gardening episode on rooftop gardening to be aired on The Life Channel in 1997.

Improving access to the site is one of the objectives of the management at the warehouse, and would be an excellent way of realising more of the garden's potential benefits.

Waste Reuse

All materials used in the garden were salvaged from the urban waste stream, and I was therefore quite successful in meeting my goal of making a productive use of neglected resources. The garden beds were salvaged from broken packing skids. The soil is municipal compost. The boxes and plastic lining the beds were taken from the refuse produced by Field to Table. Discarded electrical conduit served as supports for runner beans and tomatoes (Photo 3.5). Old pickle buckets provided planters for tomatoes, tomatillos, and eggplants. Bushel baskets grew cucumbers and summer squash. And shade was provided by corn crates. The only hardware purchase I made was for screws to hold the garden beds together.

Costs & Revenues


There are two types of monetary cost associated with the garden. The first is for one-time expenses (Table 3.2), and includes the tools and materials needed to build the garden infrastructure, and provide for its maintenance (the cost of the items marked with an asterisk is an estimate as they were already at the warehouse). The second set of costs is ongoing, and includes materials for making soil blocks, and the fertilisers required to maintain nutrient levels in the garden beds (Table 3.3).

The first of these costs is spread over the life of the garden, and in this case, represents a very small expense for a business start-up. The second set of costs could almost be eliminated by using broadcast sowing for the main crop (rather than using soil blocks), and by using compost to maintain fertility.

Table 3.2 One-time costs
Item Function Cost
Soil blocker Making soil blocks $34.95
Watering mats Watering soil blocks $39.90
Hand saw Cutting packing skids $24.00
Screws Assembling garden beds $20.00
Shovel* Moving compost and soil mix $19.99
Rake Preparing garden beds $19.99
Trowel* Transplanting soil blocks $7.95
Hose* Watering $50.00
Hose nozzle Watering $4.58

TOTAL $221.36

Table 3.3 Ongoing costs
Item Function Cost
Blood meal (3 x 2 kg) Source of nitrogen $20.66
Kelp meal (5 x 2 kg) Provides micro-nutrients and potassium $34.44
Bone meal (1 x 2 kg) Source of phosphorus $3.44
Lime (1 x 15 kg) Used to adjust soil pH (acidity) $3.99
Black peat (3 x 30 l) Blocking mix component $10.32
Brown peat (1 x 4 ft3) Blocking mix component $6.89
Potting soil (4 x 20 l) Blocking mix component $13.75
Sand (30 l) Blocking mix component $8.60
Vermiculite (6 x 4l) Blocking mix component $9.59
Zip-closure bags (2 x 30) Packaging salad mix $4.58

TOTAL $246.26

Labour should be considered when accounting costs, and indeed I believe that it this cost that will determine the economic viability of any rooftop gardening enterprise. Labour, like monetary cost, can be divided into one-time and ongoing investments. I estimate that I spent about 85 hours getting the site and work area prepared. This would have been doubled or tripled if I had had to move the compost to the roof. As for weekly labour, I estimate that about 10 to 12 hours are required per week. This includes the time required for watering, weeding, harvesting & washing the salad mix, blocking, transplanting, and general cultivation and maintenance.


Revenues from the project were not nearly what I had expected. As mentioned earlier, lettuce yields were far lower than I anticipated. There is also a difference between selling wholesale (as I did to Field to Table and Juice for Life), and selling retail.

My revenues amounted to approximately $24 per week. This is from the sale of salad mix at a wholesale price of $4 per pound (occasionally I sold lettuce only, at $3 per pound). This is a fairly low price, but it is the same as the cost of similar salad mix from California. Over the course of the 15 weeks I expect the garden to be in production, this will amount to a total revenue of approximately $360. This is enough to cover the ongoing expenses and most of the start-up cost, but leaves nothing for the labour invested in the project. This is still a respectable yield, however. According to Lee (1993), a market garden that yields $1 per square foot is doing well. Only 12 of my beds (288 square feet) are devoted to salad production, so I am producing $1.25 per square foot.

If the salad mix were sold at retail, revenues would be much higher. Mesclun from a local organic grower retails at $5 per 100 g (about $22.50 per pound). If I had a retail venue, revenues from the garden could have exceeded $2,000 at this price (the mix is of comparable quality). This is enough to cover all of the production costs, and provide a fair return on the 10 to 12 hours of labour required each week.

I believe that with a few improvements to the cropping plan I initially devised, these revenues could be increased further. This could be done by increasing the initial planting density, harvesting lettuces as cut-and-come-again greens, and devoting more space to the salad mix by eliminating some of the more inefficient crops. There is also the potential to extend the season by a few weeks in both the spring and fall with the use of glass frames on the garden beds. Outlined below in Tables 3.4 and 3.5 are the crops which I would recommend for future cultivation at this site and those which I would neglect. The tables are arranged in the following order: salad greens, edible flowers, herbs, and finally vegetables. (It is important to note that my recommendations are for this site, elsewhere, and indeed in another season, it may be found that the problems I experienced do not materialise, or that problems I did not encounter are significant. Advice should always be taken with a grain of salt.)

Table 3.4: Crops recommended for cultivation on the rooftop
Plant Comments
Arugula A fast growing and popular salad green. Requires some shading during hot weather to germinate and grow well.
Cress I'm still of two minds regarding this green. It did quite well during cool weather, but was a complete disaster when the temperature rose. Yields are also fairly small, and it was the first green to decay in the refrigerated salad mix. It does have a nice flavour however, and perhaps more attention to its cultivation would produce better, more dependable yields.
Mizuna (Mustard) A strong grower in cool weather. Produces heavy yields. Mild flavour allows it to make a significant contribution to the salad mix. Very beautiful leaves. Requires protection from flea beetles. Cool season only.
Osaka Purple Mustard Stronger flavour than mizuna, but just as vigorous. Attractive addition to the mix, but must be used sparingly. Cool season only.
Tatsoi (Mustard) Thick, spoon-shaped leaves generated much interest at Field to Table. It is mild and can therefore form a large part of the mix. Cool season only.
Lettuce Cool season varieties: Black Seeded Simpson, Red Sails, Rossa d'Amerique, Rossa di Trento, Valeria
Lettuce Hot season varieties: Buttercrunch, Lollo Rossa, Red Riding Hood, Red and Green Salad Bowl, Sierra
Malabar Spinach Could prove useful, but must be started indoors very early (around 10 weeks before the last frost). Hot season.
New Zealand Spinach Same as Malabar spinach. See figure 3.1 for proper seed treatment. Hot season.
Purslane Very dependable hot season green. Succulent leaves are delicate, and require separate washing and drying.
Borage Beautiful blue flowers make a nice addition to the salad mix. Plants produced acceptably in 3" beds, but not as healthy as those grown in deeper soil. Blooms in June.
Calendula Greens are not edible, but flower petals are a spectacular addition to the salad mix. Touch of Red is an attractive variety. Blooms in July if planted outdoors at last frost (could be started indoors for earlier yield).
Fennel Flowers a useful addition to the mix if harvested young. Too slow growing to be of much use as a rooftop garden vegetable.
Johnny-Jump-Up Viola tricolour. Small plants require very little space and produce many colourful, edible flowers. Seeds require cold treatment to germinate well. Flowers in June from direct seeding.
Shungiku Also known as garland chrysanthemum. Young greens are edible (they have a strong celery flavour, and should be used sparingly). Flowers produced in late July from direct seeding.
Epazote Mexican herb considered indispensable for flavouring beans. Apparently unavailable in Toronto, this herb deserves some market research.
Purple Perilla (Shiso) A few plants are all that is needed (four served me well). The large cumin flavoured leaves make a nice garnish if sliced thinly. (They must be sliced thinly, as their flavour is quite strong).
Tomatillo Like epazote, another crop that deserves some market research. These were the most outstanding plant in the garden. They grew in 18 l pails, and were extremely vigorous.
Tomatoes, Eggplant, Cucumbers All of these can be grown in pails, and could make an interesting addition to one's produce. They would be difficult to incorporate into a salad mix, however, and would most likely be better sold separately. Cucumbers are an interesting option for niche exploitation, as they can be grown along walls or over the exposed areas of the rooftop (the early season was too cold for my cucumbers to do well this year).

Table 3.5: Crops not recommend for cultivation on the rooftop
Plant Comments
Chard Although the leaves of red-veined Swiss chard are beautiful, the plant did not grow well in 3" beds. It survived, but was stunted.
Claytonia Another attractive salad green, claytonia produces very small yields for the amount of space it requires.
Dandelion Too slow germinating and growing to be useful in the rotation I planned.
Mache Same problems as dandelion.
Lettuce The following varieties did not prove acceptably vigorous: Biondo Lisce, Curly Oakleaf, Lollo Biondo.
Radicchio The entire crop of radicchio bolted before mid-July. I speculate that the problem was soil depth, as moisture and temperature until that time were both acceptable.
Agastache Also called anise-hyssop, this herb produces attractive edible flowers. It was too slow growing however to be of any use this season.
Nasturtium Both the flowers and leaves of this plant are edible, but the plants did not grow well in the shallow beds. They displayed a marked loss of vigour when compared to normal plants.
Chervil This anise flavoured herb would make a nice addition to a salad mix, but it too failed to grow well in shallow beds.
Filet Bean These beans grew surprisingly well, but the amount of space required to produce a respectable yield is prohibitive.
Kohlrabi A vigorous cool season vegetable, kohlrabi grew very well in the shallow beds. The problem with the crop is spatial; it produces only four stems per square foot, and a large space would be required to grow any useful quantity.
Okra The rooftop plants did not have the vigour normally associated with okra, but this may have been partially due to the cool spring. Like the filet bean, it requires too much space to be a useful rooftop garden plant.

Chapter 2 (The Site and the Plan)
Chapter 3 (Observations)
Chapter 4 (Conclusion), and References

pointer Return to Contents' Page pointer

Revised February 24, 1997

Published by City Farmer
Canada's Office of Urban Agriculture