environmentalCalculations.com        environmentalCalculations.com logo
Why Strive to Minimize Environmental Impact?
To many, especially likely visitors of an environmental site such as this one, this page's title question may seem obvious. For those, this entry is entirely optional, it may only reaffirm already deeply held convictions, but I am fairly sure it contains some useful, possibly new, information. Others may have stumbled on this page by chance, but are convinced that there is no reason to go out of your way to achieve any environmental goals, and perhaps that if an environmental problem becomes serious enough, market forces will modify relevant costs so as to discourage people, governments and businesses from further contributing to the problem. If you fall into this category, I encourage you to continue reading, because while this view is entirely logical, there is mounting compelling evidence it does not work in reality, at least not for current environmental matters. At the same time, I can't swear you will end up convinced. Others still may be on the fence, but are open to persuasive, cogent arguments in favor of, or against, voluntary environmental impact minimization actions. If you fall into this latter category, I definitely hope you will keep on reading.
What works poorly: running out of resources

Proven world natural gas reserves, 1980-2007 Proven world oil reserves, 1980-2007
Figure 1: Proven global reserves (dashed) and consumption (solid) of natural gas (left) and oil (right), 1980-2007. Both panels are from Shafiee, S. and E. Topal, 2009: When will fossil fuel reserves be diminished? Energy Policy, volume 37(1), 181-189.

Let's start with an often invoked motivation for impact minimization efforts I find logically uncompelling: pending catastrophic environmental outcomes stemming directly from resource shortages. While well-informed, cogent versions of this argument, e.g., this, this or this, are definitely worth considering, I still find them partially persuasive at best, offering a weak model of reality. But what I mostly argue against below is less nuanced, resource scarcity literalism (see, e.g., this).

Limiting depth of oil & gas deep ocean prospecting by year, from http://rogerpielkejr.blogspot.com/2010/04/what-technology-can-do.html The reason that argument has not often carried the day—and I doubt it will very often in the future—is innovation; when availability of a given resource declines appreciably, but well before it becomes materially scarce, market motivated research and technical innovation are focused on improving methods of locating and extracting the resource, or in some cases into identifying abundant alternatives for hard to get resources, restoring the availability-to-demand ratio of the original resource. For example, consider this brilliant example of obviating the need for hard to get, polluting, heavy metals with, well, dirt. Or take natural gas and oil proven global reserves (shown in Figure 1 for 1980-2007); have these reserves declined in recent decades?! Not really, despite rapidly rising cumulative consumption. (But it is true that the rate of supply increase has now leveled off, and will begin declining in the not too distant future.) How is that possible?! Improved extraction technologies (for example, the deep ocean—now a major fossil energy source, as this paragraph's upper right figure makes clear, made popularly infamous by the Deepwater Horizon explosion—was all but closed for oil exploration until a couple of decades ago).

Scaled global metal reserves, 1955-1993
Figure 2: Estimated lifetime—essentially the ratio of global consumption to known reserves—of the the industrially important metals zinc (Zn), lead (Pb) and copper (Cu). From Wellmer, F.-W. and U. Berner, 1997: Factors useful for predicting future mineral-commodity supply trends. Geol. Rundsch (Geologiska Foreningens I Stockholm Forhandlingar), published by Springer-Verlag, volume 86, 311-321.

Is this behavior unique? some inexplicable, irreproducible fluke of fossil fuels? No, not really. Take, for instance, global reserves of the three industrially important metals figure 2 addresses. Over roughly four decades starting in the mid-'50s, the ratios of those reserves and their respective global demands have not really changed appreciably, and at times have actually gone up. Figure 2 does not have the telltale appearance of a pending scarcity catastrophe, even though during this time actual cumulative demand of these metals increased dramatically. Clearly, the rate at which we discover new deposits of these metals has roughly kept pace with rising demands.

Are there limitations to this view? Are some resources hitting the immovable wall of resource finality applies? To some extent there are. Take, e.g., the aspects of global agricultural production figure 3 summarizes, portraying a mixed, nuanced picture. Over the shown roughly half century, world arable land as a percent of global land area has expanded steadily, at least through 1990 (figure 3a), but only by about 10% at most. At the same time, world population (b) has more than doubled. As a result of the arable land expansion falling well short of population growth, per capita arable land area (c) declined by 45% (the 2009 value is 55% of 1961's). At the same period, however, per acre productivity of cereals and other crops (3d) has gone way up. We can derive an overall crop production index per capita, by dividing panel 3d's green curve, per acre crop productivity, by population, 3b's blue curve. This scaled index (3e)—arguably the most pertinent measure of food supply adequacy, at least in terms of average food quantity—has increased by more than 40% between 1961 and 2009.

Some characteristics of the global food supply system
Figure 3: Panels a and c: arable land as percent of global land area (a), and hectare (ha, or 104 m2) per person (c). Panel b (blue): global population, billions. Panel d: land crop productivity. In magenta (left vertical axis) is the global mean annual cereal yield in metric tons per hectare, and in green [right vertical axis, with (indicated) divisible by 20 values shown in dashed] is a dimensionless index of overall global mean crop production, where 100 corresponds to the global production in the year 2000. Panel e shows the ratio of crop production index (d, green) and population (b, blue), an index of per capita crop production. Throughout, round and mid decades are indicated by open circles and solid squares. All data are from the World Bank.

While the significant growth in per capita crop production over the last half century is consistent with my contention that hitting the wall of actual material scarcity is rarely realized, the decline in per capita arable land is not. The apparent self-contradictory duality of the food production example showcases the current food system's main triumph, seemingly inexorably rising supply of food calories, while also implicitly emphasizing (as I demonstrate below) its equally important shortcoming, the large and rising environmental costs the supply increases have exacted. This dualism also highlights what I view as the key failing of the "hitting the scarcity wall" argument:

by repeatedly sounding the alarm about pending calamities that never materialize, and by systematically failing to explain unfolding reality even with the benefit of hindsight, "hitting the scarcity wall"— a low hanging fruit for partially deserved ridicule—unwittingly undermines environmental efforts by diverting attention from what actually matters, the rising environmental costs of dwindling but not yet scarce resources.

To appreciate the dilemma, consider the environmental costs the expanding caloric supply has exacted. While this expansion is partially attributable to agricultural land expansion (figure 3a, and see also further down), its central element is rising per acre productivity (figure 3d). In turn, the two key ingredients of rising per acre productivity are irrigation and fertilizer, both with major environmental consequences. Let's examine irrigation withdrawals first. While figure 4a shows a factor of 5 or so rise in Irrigation water withdrawal (IWW, blue), arguably more remarkable is the close tracking of global total and irrigation related consumptive withdrawals (Gwc and IWC, respectively, essentially the net water losses). What this close tracking means is that the lion's share of the rise in global water consumption is attributable to irrigation, which is of course consistent with the (logistic-like) rise in both crop- and pasturelands figure 4b shows. Note that consumptive water use comprises all water uses that directly compete with (and sometimes rob) other uses. For example, consumptive use for irrigation in California's central Valley directly competes with, and undermines, ecological integrity of Western aquatic ecosystems and species. This often amounts to a simple zero-sum choice between agriculture and other uses, such as, e.g., the choice between the water needs of farmers, Native Americans, and wild salmon in Oregon's Klamath Basin or the Columbia. So it is not running out of water—a geophysical virtual impossibility on large scales—we should be concerned with, but the rising environmental costs of exploiting ever more oversubscribed existing supplies.

Another aspect of the expanding irrigation fueling the steadily rising per acre productivity is the need for dams. When irrigation is the deciding factor (figure 4c's green dots), the dams are needed in order to regulate major river flows so as to maximize water extraction capacity when withdrawal (i.e, in many cases, primarily irrigation) needs dictate, not when the rivers naturally crest. Dams, a dominant continental scale global presence (figure 4c), raise many serious environmental issues. This is too broad a topic to address here in any meaningful way (you can read a nice review of dam hydrology here and a discussion of the benefits, as well as a few short-term pitfalls, of dam removal here), but suffice is to say that globally proliferating dams and the reservoirs upstream of them (figure 4d shows a roughly 12-fold increase in global storage capacity upstream of dams) are a key force behind human alteration of landscapes, the hydrological cycle on land, biodiversity declines, and species endangerment, among many other undesirable impacts. Again, it is not running out rivers to dam that is most concerning, but the environmental burdens existing and planned dams exert.

Aspects of global irrigation
Figure 4: Some environmental costs of expanding irrigation for raising crop yields. A: irrigation water withdrawal (IWW) and consumption (IWC) relative to global water withdrawal (GWW) and consumption (GWC). Water withdrawal is diverted surface water or groundwater, of which only a fraction is lost by evaporation or to humans or livestock use. B: the contemporaneous rise in global cropland and pasture lands, in 106 ha. C: global dam distribution by principal use; of interest here are dams marked by green dots, whose main use is irrigation. D: 20th century global reservoir storage capacity history. E: the relatively modest-sized Army Corps of Engineers' Lower Granite Dam, on the Snake River's mile 107, not far from Walla Walla, WA, obtained from the Corps' Digital Visual Library. Panels a-b are from Scanlon et al., 2007, Water Resources Res., 43, W03437. Panels c-d are from Biemans et al., 2011, Water Resources Res., 47, W03509.

The second element of agriculture's rising per acre output is fertilization. In this issue too, the scope of the problem and the volume of literature documenting nitrogen fertilizer's many adverse environmental impacts far exceed the scope of this modest venue (this, this or this are nice lay-person introductions, while a classic scientific introduction is here). Below I therefore summarize briefly a few key issues. Here again, the point is that the main issue with food supply is not so much running out of land—we can easily double the number of adequately but not overly nourished people globally over a year or two timescale with significant but tractable structural changes to the food production system—but the rising environmental costs of getting more out of existing agricultural land while failing to reconsider priorities, customs, preferences and resource allocation patterns. Graphically, the problem is the logistic nature of many resource extraction histories, such as those shown in figure 4a, b or d: a rapidly (quasi-exponentially, if you want to get technical) rising phase (e.g., 1930-1975 in figure 4a or 1950-1975 in figure 4d), followed by declining rates (during which the curves gradually level off, approaching horizontal lines, e.g., after 1985 in figure 4d). This leveling off indicates that we still need more, but manage to add less and less as time goes on; that's precisely the time for innovation to kick in, to displace demand toward more available resources.

A life saver in some circumstances and a driving force behind the Green Revolution (see this or this), overuse of nitrogen fertilization comes at staggering environmental costs. To begin with, its production uses tremendous amounts of fossil fuels (both for energy and as a raw material, the latter alone accounting for 4% of global natural gas consumption), contributing about 1.2% to global greenhouse gas emissions, and to air pollution. The trick here is that the source of all N used for fertilizer is the atmosphere, which is almost 80% N. But nitrogen in the atmosphere is in the form of molecular nitrogen, N, in which the two participating atoms are bound together with a triple bond (they "share 3 electrones" Wikipedia has a nice entry on N), which is incredibly hard to break, but must be broken in order to render largely biologically inert molecular atmospheric nitrogen biologically active and available. Naturally (mostly be bacteria), about 100 million metric tons (1011 kg) of inert atmospheric nitrogen is converted into biologically active forms ("fixed") per year. In recent years, at least as much—probably about 120%—nitrogen is fixed annually by industrial processes, nearly all for fertilizer production.

By far most troublesome is nitrogen fertilizer's central role in undermining water quality in rivers, ponds and lakes, and the world coastal oceans. In a nutshell, the issue is as follows. Nutrients—elements, primarily nitrogen and phosphorus, needed for basic life functions such as protein synthesis; think about nutrients as "food" plants need—in most natural, undisturbed aquatic systems (e.g., pristine ponds or lakes) are fully or nearly used up by plants (algae), so that the nutrient concentrations in the upper water column are nearly zero. Then, the system "productivity"—how much carbon in the form of atmospheric CO2 the system transforms, or fixes, into living tissue per unit time—is maximized, and any further productivity enhancement is limited by nutrient availability. If additional nutrients are somehow introduced into the system—by fertilizer rich runoff in most cases—nutrient availability limitations on productivity are at least temporarily removed, and productivity rises. Think about wading into a warm pond in August through a thick, green slimy mat of aquatic vegetation, and you get the picture, or you can go here for breath taking photos. These events are called algal "blooms", and you can see an example below (figure 5).

Red tide in La Jolla, CA, Aug. 2005. Photo by A. Diaz.
Figure 5: Red tide off the coast of La Jolla (essentially San Diego, CA) during August 2005. Photo by A. Diaz.

In coastal oceans and estuaries the world over (see this compelling map of global distribution), algal blooms occur, on a grand scale, typically in summer. Some (e.g., in the cold coastal waters along northern California, Oregon and Washington) are perfectly natural. Many, however, are due to fertilizer rich runoff from agricultural intensive lands (the one in the northern Gulf of Mexico off the mouth of the Mississippi is particularly notorious and well-documented). Figure 6 presents some aspect of dead zone mechanism and fishery consequences.

schematic of ocean denisty variability with depth The coastal ocean typically comprises, in a highly idealized manner, two layers (right). In the top layer (spanning the surface to a depth, highly variable in space and time, ranging from a few, to tens, or even hundreds or more meters below the surface), sea water is typically fairly warm and fresh, and thus light. Below the top layer is a thicker layer containing heavier, colder, often saltier water . Figure 6's schematic shows these two layers, separated by the thick cyan line, the so-called pycnocline (a thin layer, marked "rapid transition" in the upper right figure, in which water density increases rapidly with depth, idealized here as a single line). Because this configuration is stable (lighter water rides over denser water), the two stacked water masses exchange heat and other properties sluggishly (vertical mixing is slow). Near the surface, where in summer water is intensely heated by the sun and is thus warmer and more buoyant, there is plenty of sunlight for algal photosynthesis. Algal growth is thus so ubiquitous, using up nearly all the nutrients in the surrounding water. The importance of the pycnocline is that it prevents replenishing upper layer nutrients by deeper, nutrient rich water, because raising the lower layer's heavier water required energy. Thus under normal conditions, algae growth near the ocean surface is limited by nutrient availability. But if you add a strong flux of nutrient rich water from a river draining a large agricultural province (e.g., the Mississippi), the upper left thick arrow into the figure 6's "ocean", you lift this limitation. The result is an algal bloom near the surface, shown as the greenish upper layer. There, the water is very rich with oxygen both because of the proximity to the atmosphere and because of internally produced O2 due to the algal photosynthesis. The blooming algae live briefly, and die quickly. Some of this dead organic matter makes its way to the bottom, where it behaves just like that head of lettuce forgotten in the back of your refrigerator; it decomposes, or decays. Organic matter decomposition—the reverse of photosynthesis—requires rather than generates oxygen. The needed O2 is supplied by the water column, and the result is O2 depleted water column. The oversubscribed oxygen supplies of the near bottom waters could have been replenished by downward oxygen flux from surface waters that are oxygen saturated because of their intimate contact with the overlaying atmosphere, but this flux is rendered too sluggish by the afore-mentioned stratification—near surface waters less dense, or more buoyant, than deeper, denser, waters—to suffice. This is hypoxia (low dissolved oxygen) or, if it's even more intense, anoxia; either means insufficient levels of dissolved oxygen in the water.

And marine life forms? Well, they need dissolved oxygen, without which they suffocate. Large fishes in small, mild hypoxic regions can swim away. Shellfish and other slow- or non-moving creatures of the sea, on the other hand, die, just like the crab figure 6 depicts, posthumously. Figure 6's lower left inset shows the history of effort required for Gulf of Mexico shrimping; clearly the bottom area that needs to be scoured for a fixed shrimp catch increases with passing years, with every decade's average catch per ha lower than the previous decade's average.

Some basic dead zone physics
Figure 6: Elementary dead zone principles and consequences. The main schematic depicts basic coastal ocean physical structure relevant to the dead zone mechanism, where ρ, S and O2 denote seawater density, salinity and dissolved oxygen, and up/down arrows denote elevated/suppressed levels of the variable to their left See text for further details. The upper left inset shows a dead crab, characteristic of the fate of shellfish in reduced dissolved oxygen areas such as dead zones. The lower left inset, courtesy of Dr. James Nance of the National Marine Fisheries Service's Galveston Laboratory, shows normalized Gulf shrimp catch in kg shrimp caught per unit effort, which is defined as scouring an ocean bottom area 1 ha in size.

I hope the above synopsis' brevity doesn't obscure the scope, magnitude and seriousness of reactive nitrogen (biologically available nitrogen not locked in the largely inert N2 molecule) impacts on aquatic environments. Given the severity of the problem, the high environmental costs of fertilization, you probably envision a global trend underway toward ever increasing fertilization efficiency, toward getting more out of less. As figure 7 makes clear, just the reverse is true, with decreasing yields per unit N applied in all continental agroecological systems. For example, in the U.S., grain mass produced per N mass applied declined from roughly 90 in 1960 to about 30 in recent years. Further, the longer a continental agroecological system has been subject to routine, widespread fertilization, the less responsive its yield is too added nitrogen, with western Europe getting the least grain, followed by the U.S., culminated by South America, whose industrial fertilization history is, on average, much shorter. Here again, the issue is not so much running out of land to feed the world, but rather the ostentatiously wasteful way we go about using available land.

Global nitorgen fertilization efficiency
Figure 7: Global nitrogen fertilizer efficiency (grain production per unit nitrogen fertilizer), from Hatfield and Prueger, 2004, ISBN: 1 920842 20 9.

As I mentioned earlier, a less impactful, yet very environmentally costly, element of rising agricultural productivity (figure 3e) is expanded agricultural land. While figure 3a quantifies the process globally, figure 8a breaks it down regionally. In the developed world (e.g., North America here; Europe—not shown—behaves similarly until 1990, but features a massive jump following the collapse of the Soviet Union), agricultural lands have either held steady or declined slightly. Conversely, in the developing world (e.g., figure 8a's uppermost three curves), agricultural land extent steadily rises. The important issue here again is the environmental costs of this rise, mostly through tropical deforestation.

Some regional aspects of global agricultural land
Figure 8: Some regional aspects of global agricultural land. A: annual time series (historic evolution) of agricultural land spatial extent in six global regions, from Google Public Data, based on World Bank data. B: fraction of cropland (arable land and permanent cultures) in each 5' × 5' gridcell, from Social Ecology Vienna (Erb et al., 2007, J. Land Use Sci., 2(3), 191-224).

Tropical deforestation for cattle grazing in Praguay's Gran Chaco
Global deforestation in the 20th century's final decade
Global deforestation in the 20th century's final decade
Figure 9: The up-close-and-personal tragic face of tropical deforestation. Left: Paraguay's semi-arid Chaco, from Wikipedia. Lower right: a felled forest, from flickr, photo by crustmania. Upper right: geographical distribution of global forest losses, in percent of existing forest, in the 20th's century final decade (using again World Bank data and graphics), with red, orange and yellow indicating, respectively, >1%, 0.5-0.9%, and 0.1-0.4% loss over the considered decade.
A quasi-theoretical perspective

If you are a Hubbert Peak (HP) aficionado, you may appreciate the following HP-inspired interpretation of my point. Just to get everybody on the same page, let's review briefly some necessary HP background.

Suppose you are interested in a given resource, whose cumulative production rate pc is given by figure 10's equation 1, where pmax gives the total amount of the resource known or thought to exist. Note that while at any given time pmax is thought to be a given constant, technological advances may require updating its value periodically. The larger α is, the smaller the initial cumulative production rate as a fraction of pmax is, and the larger β is, the faster the resource is exhausted. The rate at which the resource is produced at any given time—the instantaneous production rate—is the time derivative of pc(t), p(t), given by figure 10's equation 2. Figure 10a shows pc(t) for pmax = [100, 130, 160, 190, 220], with α = 80 and β = 0.02 in all 5 cases, in red, green blue, cyan and magenta, respectively. Similarly, figure 10b shows pc(t) for the same 5 cases.

The real world relevance of this is that equation 1, and thus figure 10a's curves, describe very well availability of most resources on which humans depend. When a resource first becomes crucially relevant to humans (e.g., coal immediately following Watt's introduction of the steam engine, or oil shortly after cars became widely available), its exploitation or extraction rises rapidly (exponentially), e.g., between t = 40 and t = 180 in figure 10a. After a while, however, most of the low hanging fruits have been picked, so that continued extraction or exploitation comes at an increasingly rising cost (e.g., while the first oil wells were only vertical and tapped land reservoirs just beneath the surface, today's reservoirs can be some 10 km beneath the bottom of a 3 km deep ocean, and can comprise equally long horizontal segments).

Now consider the hypothetical scenario in which the instantaneous production is 0.5 (the horizontal black line in figure 10b). Imagine this instantaneous production rate is first achieved when your best estimate of the total resource available is pmax = 100 (shown in red). From an overall availability standpoint, this instantaneous production rate is realized where figure 10a's black curve intersects that panel's red curve. (For ease of visualization, figure 10c reproduces a magnified version of figure 10a's gray shaded region.) Next, suppose technological innovation pushes your best estimate of total resource availability to pmax = 130 (shown in green). Following the same logic, from an overall availability standpoint, this instantaneous production rate is now (after pmax was updated from 100 to 130) realized where figure 10a's black curve intersects that panel's green curve; clearly earlier relative to total availability. Follow the same argument until pmax = 220 (magenta), traversing figure 10c along the black curve from upper right to lower left. Because as pmax increases along [100,220] (corresponding graphically to the red→green→blue→cyan→magenta progression) the black curve's crossing of the cumulative production curves occurs earlier relative to total availability, the percent exhausted at that point gets smaller, as figure 10d makes clear.

With the HP perspective, innovation is the force moving cumulative production curves to the left, thereby yielding earlier relative times at which a given instantaneous production rate is realized, and thus smaller corresponding exhaustion percent. This is why the resource scarcity argument is, in my view, a weak one.

Peak perspective on the scarcity wall.
Figure 10: Hubbert Peak perspective on the scarcity wall. In all cases shown here α = 80 and β = 0.02. Maximum cumulative production pmax [equation (1) on the right] varies from 100 (red) to 220 (magenta), in increments of 30. Panel a shows cumulative production [pc(t), equation (1) on the right] for the five pmax values. Panel b presents the time (t) derivatives of the five pc(t) curves, given by equation 2 on the right. Still in panel b, the horizontal black line marks the maximum of the red derivative curve [pc(t) with pmax = 100], 0.5.

So then why minimize impacts?

Several coupled key reasons.

1. Financial savings: This is an obvious one; most optimization measures save money, and some save lots of money.

2. Duty: Today, most traditional resources and some emergent ones (e.g., various precious metals needed for modern IT) are at the declining phase of their availability, like figure 10a's curves after, say, t = 280. They are not yet scarce in any simple meaning of the word, but they sure are getting more environmentally expensive to extract and exploit. At that point, when resource additions are slow and declining, availability becomes almost a zero-sum game, with use by one person meaning reduced availability to others, so that continued unchecked resource use by one directly undermines the well being of others. This to me is arguably the most compelling reason to minimize impact. At this point, it is very tempting to invoke no lesser authority than Immanuel Kant, whose moral duty idea, especially his first formulation of the categorical imperative, is extremely pertinent:

''Act only according to that maxim whereby you can at the same time will that it should become a universal law without contradiction'' Kant, I., Grounding for the Metaphysics of Morals, translated by J. W. Ellington, 3rd Edition, Hackett June 1st 1993, 78 pp.
Following this adage, each consumer must ask: Can I generalize my own consumption into a universal rule whereby everybody consumes at this rate? At the declining phase of resource availability, where zero sum prevails, the answer is no. According to Kant, it is therefore morally imperative that we conserve, or minimize impact.

3. Elegance: Think of, e.g., Lance Armstrong's biking, Marlon Brando's acting, Fred Astaire's dancing, or Hemingway's writing. As I see it, what sets them apart, what makes them great, is economy. Having studied their disparate crafts to perfection, they eliminated all that's not essential, and nothing more, until they were left with the irreducible essence of their crafts. Perfection, that's easily identifiable even for those not particularly interested in the specific craft.

This observation also applies to most any human-environment interaction. Virtually any resource use can be made more economic, parsimonious, efficient; virtually anything we do involving resource use can be achieved with less, sometime dramatically less. If it is possible to build a financially tractable, attractive house whose operation requires no fossil fuels, and that even over its full life cycle (including construction and manufacturing of all elements) uses dramatically less fossil fuels (e.g., this or this claim to—and might—do just that), why wouldn't you?! If you can get from a to b by car using a fraction of the gas, why wouldn't you?!

4. The health connection: This addresses personal benefits, in departure from the benevolence of the first and the abstraction of the second. Much of modern resource use replaces the use of human muscles; driving instead of walking, elevator riding instead of climbing stairs, leaf blowing instead of raking, to name but a few random examples. Yet the human body has evolved for muscle work; the evidence documenting its rapid functional degradation and declining quality of life is overwhelming. There is thus a direct parallel in many aspects of resource use: the less you use, the healthier you are.

Last modified on GMT by Gidon Eshel