Dataset: Broadbalk mean long-term yields of winter wheat 1852-2018

Citation:  Sarah Perryman, Margaret Glendining, Andy Gregory, Paul Poulton (2023). Broadbalk mean long-term yields of winter wheat 1852-2018 Electronic Rothamsted Archive, Rothamsted Research 10.23637/rbk1/meanWWYields1852-2018-02
to RefMan to EndNote

Figure: mean long-term winter wheat grain yields 1852-2018

Figure: mean long-term winter wheat grain yields 1852-2018


This dataset provides mean long-term yields of winter wheat (grain at 85% dry matter) from 1852-2018 in response to selected treatments on the Broadbalk "Classical" long-term experiment. The Figure shows the long-term trends and yields achieved under various scenarios in comparison with the original practices at the start of the experiment. The data are mean values over various numbers of years and/or cultivars; there is considerable year-to-year variability (data for individual years are available from the Broadbalk Complete Open Access dataset or e-eRAdata). Trends reflect the treatments and agronomic practices introduced on Broadbalk such as continuous wheat (solid lines) and rotations (dashed lines), old varieties until 1967 versus modern short-strawed cultivars since 1968, and the management of pests, weeds and diseases, introduced at various times since the 1920s.

This new updated version of the figure includes data from 2017 & 2018 and a treatment of "best continuous wheat" as a comparison to the "best rotational wheat" treatment - it was created for the new summary publication on Broadbalk 1968-2018, "The Broadbalk Wheat Experiment, Rothamsted, UK: Crop yields and soil changes during the last 50 years" (2024) Poulton et al. Advances in Agronomy.

The field experiment was originally set up as long strips the length of the field and weeded by hand-hoeing. That eventually became impractical and so the experiment was divided into five sections in the 1920s and regular fallowing was introduced to control weeds. The introduction of herbicides on the whole experiment (except on what is now Section 8) in 1964 removed the need for fallowing. In 1968, major changes occurred; the introduction of short-strawed varieties with greater yield potential, and the further division of sections to allow for rotations on some sections of the experiment, so allowing for the comparison of yields of wheat grown continuously or as the first wheat after a two-year break; the form of fertiliser N being tested changed and a higher rate (192 kg N ha-1) was introduced. Other changes followed; Since 1979, summer fungicides have been used, exploiting the greater grain yield potential of modern cultivars; From 1985, two further higher N rates have been tested (240 and 288 kg N ha -1). Since 2000, P has been withheld from selected plots (as plant-available P is not limiting) and split-N applications was introduced on some plots. New varieties of wheat are introduced to maintain relevance with current agricultural practice.

Since 1968 there have been various crop rotations on five sections, indicated top right of figure, and the largest yields are from the wheat in rotation rather than continuous wheat. They are usually from the first wheat crop and often from the combination of FYM + 96 kg N ha-1 (144 kg N ha-1 since 2005, plot 2.1) rather than from inorganic fertilizer alone. For continuous wheat, the best NPK treated plots yield more than either FYM alone (plot 2.2) or PKMg + 144 kg N ha-1 (plot 8). The largest annual wheat yields ever recorded on Broadbalk was 13.8 t ha-1 in 2014 (variety Crusoe; plot 17 N1+4+1PKMg). The largest yield in the drought year of 2022 was 12.22 t ha-1 (variety Zyatt; a 2nd wheat on plot 2.1 FYM+N3), the tenth-highest yield ever recorded. Note that the best yields are not always achieved with the highest N rate. The figure shows the mean best 1st wheat yields achieved from the NPK treatments, receiving up to a maximum of 288 kg N ha-1 (up to 192 kg N ha-1 from 1968-1984). This is always much more than the equivalent best continuous wheat NPK yield. Yields of continuous wheat given no fertilizer or manure (Nil plots 3) have remained at around 1 t ha-1 in general. However, these yields dipped in the late 1910s and early 1920s due to weed infestation arising (at least in part) from a reduced labour force during and after the First World War, and have again fallen since the 1980s, possibly due to lower inputs of atmospheric N.


The Broadbalk experiment started in autumn 1843, and for the first eight years, the fertilizer treatments were varied, to test specific questions. From 1852 a scheme of fertilizer treatments was established, which remained largely unaltered until 1967. The mean yields are taken from this period since 1852.

The yield values are means of every 10 years from 1852-1964, except for 1922-1934 during a transition period. After 1968, when modern short straw varieties were introduced, the means represent data for each variety grown, except for Hereward which has been split into two periods when the FYM treatment receiving fertiliser N in spring received either 96 kg N ha-1 (1996-2004) or 144 kg N ha-1 (2005-2012). Data for 2015 are not included as a spring wheat variety was grown in this years as wet weather conditions in autumn and winter delaying seed sowing.

Technical Information

Continuous wheat: Wheat grown every year since autumn 1843, except when parts of the experiment were fallowed to control weeds. The experiment was divided into 10 sections in 1968; yields shown are mean yields of Sections 1 and 9 (two of the five sections currently in "continuous" wheat and which are essentially replicates).

1st wheat in rotation: Some parts of the experiment have been in crop rotations since 1968. Yields are from the 1st wheat in rotation after two break (non-wheat) crops.

Fallowing: Between 1926 and 1967 the experiment was divided into five sections which were bare fallowed sequentially to control weeds. Fallowing was mainly in a 5-year rotation of fallow with four successive crops of wheat, with each phase present each year.

Liming: Lime (calcium carbonate, often referred to as chalk) has been applied since the 1950s to maintain soil pH at a level which does not limit yield.

Herbicides: Herbicides were introduced to the whole experiment except what is now Section 8, in 1964; previously weeds were controlled by hand-hoeing or by fallowing and cultivation.

Modern cultivars: Modern short-strawed, high-yielding cultivars since 1968. All, except cv. Apollo, were (and remain) bread-making wheats.

Fungicides: spring and summer fungicides applied as necessary since 1978, except on Section 6. 85% dry matter: By convention the yield of wheat grain is given at 85% dry matter (i.e., after drying a sub-sample to measure dry matter content). For earlier yields (pre-1954) dry matter at harvest was not measured directly but it is reasonable to assume that this would usually have been at about 85% dry matter.


Unfertilized - Nil (continuous): No fertilizer or manure applied since 1852 (plot 3).

FYM (continuous): Farmyard manure - 35 t ha -1 of Farmyard Manure (cattle) applied each year since 1843 (plot 2.2).

NPKMg with 144 kg N (continuous): PKMg + 144 kg N ha-1 applied each year since 1852 (plot 8).

FYM+144kgN since 2005 (rotational): 35 t ha -1 FYM since 1885 plus 96 kg N ha-1 1968-2004 then 144kg N since 2005.

Best yield NPKMg plots (continuous and rotational): Complete fertilizer treatment (NPKMg) giving largest yield each year (maximum 288 kg N ha -1). Since 2001, the largest yields have been selected from plots 8, 9, 12, and 15-18.

Related Documents

Related Datasets


  • Paul Poulton: Researcher
  • Margaret Glendining: Data curator
  • Nathalie Castells: Data manager
  • Andy Gregory: Project manager
  • Sarah Perryman: Data curator
  • Richard Ostler: Project leader
  • Suzanne Clark: Project member
  • Chris Hall: Data collector
  • Chris MacKay: Data collector

Dataset Access and Conditions

Rights Holder

Rothamsted Research


Creative Commons License This dataset is available under a Creative Commons Attribution Licence (4.0).

Cite this Dataset

YOU MUST CITE AS: Sarah Perryman, Margaret Glendining, Andy Gregory, Paul Poulton (2023). Dataset: Broadbalk mean long-term yields of winter wheat 1852-2018 Electronic Rothamsted Archive, Rothamsted Research

Please review our How to Credit Datasets guidance for more information.

Conditions of Use

Rothamsted relies on the integrity of users to ensure that datasets are used appropriately and Rothamsted Research receives suitable acknowledgment as being the originators of these data. Please review the Conditions of Use before downloading.

Table Of Contents

Excel file (xlsx):

Sheet 1. Figure of Broadbalk long-term mean wheat grain yields (t ha-1, at 85% dry matter) 1852-2018 for six contrasting treatments.

Sheet 2. Table of Broadbalk long-term mean wheat grain yields (t ha-1, at 85% dry matter) 1852-2018 for six contrasting treatments, providing variety names, time period of wheat means plotted, grain yields t ha-1 at 85% dry matter accompanied by explanatory notes.

This summary data is derived from annual plot data for the relevant selected plots and treatments. The original raw data is available, from the e-RA database (e-RAdata) and the Complete OA dataset [link].

This dataset is based on, and continues, an original Table, by Garner & Dyke (1969) showing yields for 1852-1967.

David Jenkinson produced an early version of the Figure in the 1970s.

In its current form it was included in the 150th Anniversary of Rothamsted publication by Johnston (1994) showing yields 1852-1990.

It was then updated by Paul Poulton and Andy MacDonald in the "Guide to the Classical and other Long-term Experiments, Datasets and Sample Archive" (2006), page 9 (showing mean yields for 1853-2000).

In 2017 this was updated further to include data to 2016 – this was the first "open access" dataset released on e-RA

This current version 2, adds another 2 years of data from 2017-2018 and another treatment, "Best continuous NPKMg" as a comparison to the "Best rotational treatment".

In accordance with the Joint Code of Practice for Research at Rothamsted, data processing in e-RA follows rigorous standard operating procedures to ensure the quality and correctness of data collected in the field through to depositing in the e-RA database. This summary data has been collated and checked by the authors; PP up to 2016; SP, MG & PP from 2017-2018.

By convention, the year is that of harvest (not sowing).

Annual yield data starts from 1852, rather than 1844 harvest (1843 sowing) because the treatments varied between 1844-1851 before settling on the long-term treatments (the complete data from 1844 are available).

We acknowledge the generations of farm staff, sample processors, statisticians, LTE managers and previous curators who have contributed to this immense data collection over many decades, some of whom are named here

Any queries, please contact email

  • The dataset Broadbalk mean long-term yields of winter wheat 1852-2018 is a published dataset from the e-RA Database. e-RA is part of the Rothamsted Long-Term Experiments - National Bioscience Research Infrastructure (RLTE-NBRI), which also covers maintenance of the Long-Term Experiments, the Rothamsted Sample Archive and Rothamsted's environmental monitoring activities including the weather stations and its role in the UK Environmental Change Network
  • The RLTE-NBRI is funded by UK Research and Innovation - Biotechnology and Biological Sciences Research Council (UKRI-BBSRC) under award BBS/E/RH/23NB0007 (2023-2028). The RLTE-NBRI is also supported by the Lawes Agricultural Trust. e-RA has been part of a National Capability since 2012, previous awards from the BBSRC were Grants BBS/E/C/00005189 (2012-2017) and BBS/E/C/000J0300 (2017-2022)
We only use analytics cookies on this site. Please refer to our Privacy and cookies policy

For further information and assistance, please contact the e-RA curators, Sarah Perryman and Margaret Glendining using the e-RA email address: