e-RA: the electronic Rothamsted Archive

Park Grass Experiment

The Park Grass experiment is the oldest experiment on permanent grassland in the world. Started by Lawes and Gilbert in 1856, its original purpose was to investigate ways of improving the yield of hay by the application of inorganic fertilizers and organic manure. Within 2-3 years it became clear that these treatments were having a dramatic effect on the species composition of what had been a uniform sward. The continuing effects of the original treatments on species diversity and on soil function, together with later tests of liming and interactions with atmospheric inputs and climate change, has meant that Park Grass has become increasingly important to ecologists, environmentalists and soil scientists.

Recent publication: Grassland biodiversity recovers: A study published in the journal Nature shows that grassland diversity on the Park Grass experiment recovers once atmospheric nitrogen pollution reduces. See Biodiversity bounces back .


The experiment was established on c.2.8 ha of parkland that had been in permanent pasture for at least 100 years. It has had three main phases:

1. 1856-1902. The uniformity of the site was assessed in the five years prior to 1856. Treatments imposed in 1856 included controls (Nil - no fertilizer or manure), and various combinations of P, K, Mg, Na, with N applied as either sodium nitrate or ammonium salts. FYM (farm yard manure) was applied to two plots but was discontinued after eight years because, when applied annually to the surface in large amounts, it had adverse effects on the sward. For 19 years the re-growth was grazed by sheep penned on individual plots but since 1875 a second harvest has been cut and removed immediately. Park Grass plan 1856-1902 (pdf). Park Grass probably never received the large applications of chalk that were often applied to arable fields in this part of England. The soil (0-23cm) probably had a pH (in water) of about 5.5 when the experiment began.

Small amounts of chalk and lime were applied to all plots in the 1880s and 1890s. The Park Grass plan of 1881 shows the chalk application in 1881 to a small part of the Northern end of the plots. In 1883 and 1887 the plots were divided into Western and Eastern halves, with lime applied to the Western half in 1883 and the Eastern half in 1887. See Park Grass chalk/lime applications 1856-1902 (pdf) for full details.

Plot nomenclature in 1856-1902 is complex. Yields are given for the whole plot, eg plot 3, 1856-1885. From 1881-1902, yields from the small area chalked in 1881 are referred to as 'C' eg 3C, and from the much larger unchalked area as 'UC' eg 3UC. Separate yields were not recorded for the limed and unlimed Western and Eastern halves in 1883 and 1887, but in some years, eg 1887 and 1889, yields are shown for the Western (W) and Eastern (E) parts of the areas chalked (C) and unchalked (UC) in 1881, eg Plots 3EU, 3WC. Plots areas are shown in e-RA, which helps with understanding which parts of the plots are referred to. For further help, please contact the e-RA Curators.

2. 1903-1964. In 1903 plots 1-13, (except 5/1, 5/2, 6 and 12) and 16 were halved and the effects of regular applications of lime (as CaCO3) were tested Park Grass chalk/lime applications 1903-1964 (pdf) . FYM, applied every four years, was re-introduced on three plots (13, 19 and 20) in 1905. In 1920, three more plots (14, 15 and 17), were halved and plots 18, 19 and 20 were divided into 3 subplots to test lime. Park Grass plan 1903-1964 (pdf)

Yields are given for the Limed (L) and Unlimed (U) halves, eg 3L and 3U.

3. 1965 onwards. In 1965 most plots were divided into four sub-plots, three of which receive chalk to maintain pHs of 7, 6 and 5 (sub-plots a, b and c respectively) Park Grass chalk applications 1965 onwards (pdf) The fourth sub-plot (sub-plot d) receives no chalk and the pH of these ranges from 3.5 to 5.7 depending on the fertilizer treatment. Dramatically different swards have evolved as a result of the different pH and nutrient status of the soils. There are 35-45 species on the unfertilised plots but only 2 or 3 species on some of the fertilised plots. From 1965 plots 5/1, 5/2 and 6 were used for microplot experiments. Plots 6a and 6b were re-included in the main experiment in 1972 but the other half of plot 6 (6c and 6d) and plots 5/1 and 5/2 have remained outside the main experiment. Since 1990, nitrogen fertilizer has been withheld from half of all sub-plots formerly receiving 96 kg N ha-1 (plots 9 and 14) as either ammonium sulphate or sodium nitrate to study processes controlling soil acidification, heavy-metal mobilisation and botanical changes. Since 1995, plot 13 has been split into 13/1 and 13/2 and FYM/Fishmeal withheld from plot 13/1. Since 1996 plot 2 has been split into 2/1 and 2/2 with plot 2/1 receiving K as potassium sulphate. Park Grass plan 1965 onwards (pdf)

In 2013 plot 7 (PKNaMg) was divided into two equal sized plots, 7/1 and 7/2. This was to test whether P fertilizer was still required, because large reserves of P have built up in the soil. No P fertilizer was applied to Plot 7/1, but K, Na and Mg applications continued. Plot 7/2 continued as before (i.e. it received PKNaMg). In addition, N applications began on plot 15; 144kg N/ha as sodium nitrate to provide a comparison with plot 11/1 which receives the same amount of N as ammonium sulphate together with P, K, Na & Mg.

From autumn 2016 all plots previously receiving 35 kg P ha-1 as triple superphosphate will receive 17 kg P ha-1 (until further notice) because of the high levels of available soil P from past P inputs. The P application to plot 20 will remain unchanged. Current K, Mg and Na rates on Park Grass remain unchanged.

Yields are given for the four sub-plots, eg 3a, 3b, 3c and 3d.

The complexity of changes even for an individual plot is spatial and temporal as illustrated by this example: Timeline for Plot 2. These adaptations were most of the original Park Grass plots were divided into two in 1903, and the effects of regular applications of lime were tested. In 1965 most plots were divided into four sub-plots, given different amounts of lime to maintain soil pH at 7, 6, 5 and an un-limed sub-plot

The botanical composition of the plots has been studied by visual surveys and by looking at the percentage contribution of the various species to the hay harvested. Chemical analyses of the crops and soils have been made and physical samples of the crops and soils preserved.

For more details, refer to the Rothamsted Guide to the Classical Experiments 2018 pages 20-30, to the Plot Descriptions and Treatments and Key References listed below.

Harvesting methods

The plots were originally cut by scythe, then by horse-drawn and then tractor-drawn mowers. The mowing maching was first used for the first cut in 1901, though it had been used for the second cut since 1881. The plots were cut each year for hay, usually in June, and a second cut taken in the autumn since 1875. A second cut was not taken every year, and not from every plot each year, if there was insufficient herbage to sample. No second cut taken 1856-1874, 1876, 1884, 1885, 1887, 1899, 1911, 1914, 1921, 1924, 1933 or 2003. In 1903-1917, the second cut was taken from the whole plot, not the Limed and Unlimed halves. The regrowth after the first cut was grazed by sheep 1856-1872, except for 1866, 1870, and 1873 and 1874, when the regrowth was mown but not removed from the plots.

Yields were originally estimated by weighing the produce from the whole plot, either as hay (1st harvest) or green crop (2nd harvest), and dry matter determined. Since 1960, yields of dry matter have been estimated from strips cut with a forage harvester. However, for the first cut the remainder of each plot is still mown and made into hay, continuing earlier management and ensuring return of seed. For the second cut, the whole of each plot is cut with a forage harvester. Consequently recorded yields of dry matter are now larger than previously as fewer losses occur. The following correction factor should be used for post-1960 data for cut 1 to give yield values equivalent to pre-1960 values:

Ycut 1 = 0.2743 x (YF 1.662 ) (Bowley et al, 2017)

where YF = yield collected by forage harvester, t/ha. The correction factor was obtained from the relationship between yields for hay and forage harvested cuts (r2 = 0.90) for a selection of plots for 1959 and 1992-1994. For more details, refer to Bowley et al, 2017 (see Key References below).

Plot Descriptions and Treatments

Data Available

All data available can be extracted using the e-RA Data Extraction Tool. Preview sets of data are also available.

Herbage yields are recorded each year. Physical samples of crops and soils have been preserved in the Rothamsted Sample Archive. For more details please contact the e-RA Curators.

With thanks to Paul Poulton and Andy Macdonald for help with compiling the text and plans.

Related Links

A selection of links to internal and external web sites refering to the Park Grass Experiment

Key References













RRes Logo - Back to RRes