CoffeariumBeta
  1. Plant
  2. Farm
  3. Process
  4. Trade
  5. Roast
  6. Brew
  7. Cup
Botany · Fruit and seed

Peaberries and bean shapes

Most coffee fruits hold two flat-sided seeds, but some hold one round seed (a peaberry), three seeds or two seeds locked in one shell. What causes each shape, how often it occurs and how it is sorted.

Across the chain: Science

Source-checkedby Coffearium verification pass· Last source-checked 22 Oct 2026· Prepared with AI assistance· 17 sources (12 tier A)· 2 contested· Updated 22 Oct 2026
Reading depth
Contents
  • A coffee fruit normally holds two seeds that face each other with their flat sides; each seed has a groove along its flat side.1,2
  • A peaberry is a single ovoid seed that forms when one ovule develops instead of the usual two.2,3
  • An elephant bean is two seeds joined together, traced to false polyembryony; when the pair separates, a shell-shaped piece (shell) is left.3,5,2
  • Reported peaberry shares of the crop lie mostly between about 5% and 10%; in an Indian cross with an interspecific hybrid parent, means were 18.54–31.18% of fruits.9,11,4
  • Peaberries are sized on slotted screens, flat beans on round holes.14,2

Two seeds, face to face#

A coffee fruit normally holds two seeds.1 They grow side by side and press against each other. Each seed therefore has a rounded back and a flat or slightly hollow face.2 A groove runs along the length of the flat face.1,2 The trade calls these seeds flat beans (chato in Brazil).2 The layers around them – skin, pulp, mucilage, parchment and silverskin – are described under coffee cherry anatomy.

Peaberries#

Sometimes a fruit holds only one seed. With no partner to press against, it grows round or egg-shaped.2 This seed is a peaberry: moca in Brazil and PB in East African grades.2,3 Colombia’s export grades, as reported to the International Coffee Organization (ICO), include a peaberry grade called Caracol.17 It still has the groove along its length.2 The East African standard defines it in one line: one ovule develops instead of the usual two.3

Three seeds, elephant beans and shells#

Some fruits have three seed chambers instead of two, and so three seeds.5 The International Trade Centre (ITC) lists three-cornered beans and describes them as semi-peaberry in character.15

An elephant bean is two seeds joined together.3,15 The East African standard calls it a genetic defect.3 A 1955 Brazilian study traced it to false polyembryony.5 According to the ITC, elephant beans are usually deformed and likely to break up in processing or roasting.15 When the joined seeds separate, a shell-shaped piece is left: the shell (concha).2 The ITC calls a piece of a broken elephant bean an ear.15

A1234B5C6D789
  1. Skin and pulp: the fruit wall around the seeds.1
  2. Parchment around each seed; it remains even when its seed fails to develop.8
  3. Flat bean: rounded back, flat face towards its partner.2
  4. Groove along the flat face.1,2
  5. Peaberry: a single ovoid seed.2,3
  6. Three-cornered bean from a three-seeded fruit.5,15
  7. Elephant bean: two seeds joined together.3,5
  8. Shell-shaped part: once separated, a shell.2
  9. Inner part of the joined pair.3
Fruits cut across. A: two flat beans. B: a peaberry. C: three seeds. D: an elephant bean beside a normal seed. Schematic, not to scale.

Why a fruit makes a peaberry#

The flower’s ovary holds two ovules, and each can become a seed.4 A 1990 Indian breeding study linked peaberries to the abortion of one ovule before fertilisation. The surviving seed then has the whole space to itself.4 Other routes to a one-seeded fruit have been reported:

  • Chromosome damage. Plants grown from irradiated seed produced more peaberries as the dose rose.6
  • Hybrids between species. Peaberries were frequent in C. canephora, C. congensis and in crosses between species.5
  • Pollination and weather. An Ethiopian paper, citing earlier Indian work, lists poor pollination, unfavourable weather and incompatible parents.10

A related fault is the empty locule: a seed chamber with no seed in it. That happens after fertilisation and is a different defect.4,7 Flowering, pollination and fruit growth have their own entries.

How common peaberries are#

Shares vary with variety and site. A 2023 Colombian study states that peaberries make up 5–7% of the harvest of arabica and robusta.11 A Brazilian trial of Mundo Novo progenies measured a slightly wider range, as the chart shows.9 Breeding material can produce far more. In a 1990 Indian study of a cross with an interspecific hybrid parent, 18.54–31.18% of fruits held a peaberry, on average per population.4 That study counted fruits, while the Brazilian trial weighed beans, so the figures are not directly comparable.4,9

0102030405060Peaberry share, %Colombian study, generalstatement(arabica and canephora)5–7%11 Mundo Novo progenies, TrêsPontas, Brazil(2007–2008)5.34–9.60%9 Devamachy, parent, India(fruits, 1984–1985)10.00–41.62% (mean 22.05%)4 S. 881, parent, India(fruits, 1984–1985)10.22–30.20% (mean 18.54%)4 F1, Selection 5, India(fruits, 1984–1985)14.59–56.43% (mean 31.18%)4 F2, Selection 5, India(fruits, 1984–1985)15.90–58.00% (mean 29.50%)4 F3, Selection 5, India(fruits, 1984–1985)12.50–50.21% (mean 28.87%)4
Data table
PointPeaberry share, %Source
Colombian study, general statement (arabica and canephora)5–7%5–711
Mundo Novo progenies, Três Pontas, Brazil (2007–2008)5.34–9.60%5.34–9.69
Devamachy, parent, India (fruits, 1984–1985)10.00–41.62% (mean 22.05%)10–41.624
S. 881, parent, India (fruits, 1984–1985)10.22–30.20% (mean 18.54%)10.22–30.24
F1, Selection 5, India (fruits, 1984–1985)14.59–56.43% (mean 31.18%)14.59–56.434
F2, Selection 5, India (fruits, 1984–1985)15.90–58.00% (mean 29.50%)15.9–584
F3, Selection 5, India (fruits, 1984–1985)12.50–50.21% (mean 28.87%)12.5–50.214
Peaberry shares reported in three sources. The Colombian figure is a general statement for arabica and canephora; the Brazilian figures are the mass on screens 10–13 for Mundo Novo progenies. The Indian figures are the range across single plants (100 fruits each, cut across), with the population mean in the label.9,11,4

Sorting peaberries out#

Peaberries and flat beans need different screens. Flat beans are sized on round holes; peaberries on slotted (oblong) holes.14,2 East Africa grades peaberries as PB: beans that pass screen 17 and stay on screen 12 (4.76 mm).3 Elephant beans form the E grade, retained on screen 21 (8.3 mm).3 Brazil sorts moca into three classes on oblong screens.2 In Brazilian grading, shells are a defect: three shells count as one.2 Screens and defect counts are explained under grading and sorting.

Ovules, locules and false polyembryony#

The coffee ovary normally holds two ovules, and the fruit two locules (seed chambers); three-locule fruits also occur.4,5 In a 1990 study of Devamachy hybrids (Selection 5) and their parents, peaberry formation was linked to space in the ovary freed by the pre-fertilisation abortion of one of the two ovules.4 Empty locules, by contrast, were attributed to abnormalities after fertilisation.4 A 1952 Brazilian study had already noted that the coffee endosperm develops late. It reasoned that the perisperm grows normally and the locule empties only later.7 The same study found that fruits picked young showed few empty locules.7 A 1966 Kenyan study likewise describes fruits in which one seed had not developed, leaving only the empty parchment.8

A 1955 study at the Instituto Agronômico, Campinas, separated true from false polyembryony. Polyembryonic seeds usually had two embryos; seeds with three normal embryos were very rare.5 False polyembryony produces the elephant bean, which can form in fruits with one, two or three developed locules.5 The study also found seeds from three-locule fruits most often in the arabica varieties purpurascens, semperflorens, Mokka and Yellow Bourbon.5

Peaberry frequency varies between and within varieties. In Mundo Novo, only some plants produced an excessive number of peaberries, and the share varied from plant to plant.7 In the 1955 survey, peaberry frequencies were variable in arabica, low in C. dewevrei and high in C. canephora, C. congensis and the interspecific hybrids arabica × C. dewevrei and C. liberica × C. dewevrei.5 Irradiating seed with neutrons raised the peaberry share linearly with dose; X-rays raised it exponentially.6 The author attributed this to chromosomal aberrations that eliminate gametes and so give one-seeded fruits.6

Stress and position. In Kenya, beans from the tip of a fruiting branch were more often abnormal than beans near its base.8 Irrigation improved bean shape and reduced the number of incompletely developed endosperms, probably through better growth in the dry season.8 According to the International Trade Centre, some cultivars produce more elephant beans, and drought or nutrition stress seems to increase the number of small ones.15

Gene expression. A 2023 study compared peaberries and flat beans from arabica plantings in Baoshan, Yunnan, China, by RNA sequencing.13 Genes for cell-wall modification, including four pectin methylesterases, were expressed many-fold lower in peaberries. The authors suggest these genes might help set bean shape.13 The study measured bean weight and dimensions and gene expression, not cup quality.13

Peaberries and empty locules in an Indian breeding cross#

A 1990 study at the Central Coffee Research Institute, India, followed Selection 5, a cross of Devamachy with S. 881.4 The authors describe Devamachy as considered a spontaneous interspecific hybrid of C. arabica and C. canephora, and S. 881 as a wild C. arabica accession from the former Anglo-Egyptian Sudan.4 They cut across 100 fruits from each of 10 plants of each parent and 25 plants of each progeny (F1, F2, F3), in October and November of 1984 and 1985, and scored peaberries and empty locules.4 Fruits with free space in the ovary are those with a peaberry or with an empty locule beside a single flat bean.4

Population (plants) Fruits with free space Fruits with a peaberry Fruits with an empty locule
Devamachy, parent (10) 31.63% (14.03–53.90) 22.05% (10.00–41.62) 9.57% (4.03–14.92)
S. 881, parent (10) 23.88% (14.80–36.40) 18.54% (10.22–30.20) 5.35% (3.20–10.68)
F1, S. 1559 (25) 60.93% (36.00–79.88) 31.18% (14.59–56.43) 29.74% (15.30–48.90)
F2, S. 2267–S. 2269 (25) 50.80% (32.56–71.42) 29.50% (15.90–58.00) 21.32% (11.21–34.09)
F3, S. 2966–S. 2988 (25) 48.49% (29.71–65.50) 28.87% (12.50–50.21) 19.61% (5.79–38.98)

Mean share of fruits, with the range across single plants in brackets; 100 fruits per plant, 1984–1985.4 By our calculation, the peaberry and empty-locule means add up to the share of fruits with free space in every population, to within 0.02 percentage points.

Before and after fertilisation. All three progenies produced more peaberries and more empty locules than either parent.4 The F1 had the most fruits with free space, but its peaberry share rose much less; the difference appeared as empty locules.4 On arcsine-transformed data, the peaberry share rose with free space in all five populations (regression coefficient 0.73–0.91). Free space explained most of the variation in the parents (r² 0.90 and 0.89) but less in the progenies (r² 0.41–0.52).4 The authors read the extra empty locules as the result of abnormalities after fertilisation, when endosperm and embryo start to develop and then stop.4 For peaberries they point to events before fertilisation: quadrivalent chromosome pairings in pollen mother cells (reported as unpublished data), which suggest translocations, and frequent pollen-tube abnormalities.4

Breeding implications. The progenies had been bred by mass pedigree selection, which the authors judged of limited use for lowering either trait.4 They propose picking F3 plants with very low peaberry and empty-locule production and checking their pollen and ovule fertility. The selected plants would be intercrossed to test for normal chromosome behaviour, and selected at the same time for high self-compatibility, before multiplication and multi-location trials.4 The lowest single F3 plants had 12.50% peaberries and 5.79% fruits with an empty locule.4 The paper proposes the scheme but reports no results from it.4

Reference values#

Parameter Value Population and basis Source
Peaberry share, general statement 5–7% Harvest of C. arabica and C. canephora Duque Dussán et al.11
Peaberry share, Selection 5 (Devamachy × S. 881) Population means 18.54–31.18%; single plants 10.00–58.00% Central Coffee Research Institute, India, 1984–1985; share of fruits, 100 fruits cut across per plant Santa Ram et al.4
Fruits with an empty locule, Selection 5 Population means 5.35–29.74% As above Santa Ram et al.4
Peaberry share, Mundo Novo progenies 5.34–9.60% (mean 7.76%) Três Pontas, Minas Gerais, Brazil, 2007–2008; mass on screens 10–13 in 300 g of hulled coffee; lowest value in the table 5.94% Raso et al.9
Locules with false polyembryony 0.21% (Typica); 4.91% (Mokka); 4.97% (Bourbon Amarelo); 5.34% (Mundo Novo) Instituto Agronômico, Brazil; share of fruit locules Alves5
Empty locules 6.60% (Typica); 29.50% (Mundo Novo); 30.00% (var. erecta) Instituto Agronômico, Brazil; share of fruit locules Alves5
Empty locules, Mundo Novo 16.6 per 100 fruits (Jaú, mean of 1950–1951); up to 112 per 100 fruits (Mococa, 1950) São Paulo, Brazil; fruits cut across and counted Carvalho et al.7
Floating (empty) fruits, Mundo Novo progenies 4.75–28.5%; below 10% judged satisfactory by breeders 100 ripe fruits in water; floaters counted Raso et al.9
PB grade, East Africa Through screen 17, retained on 12 (4.76 mm), slotted Green coffee EAS 1303
E grade, East Africa Retained on screen 21 (8.3 mm) Green coffee EAS 1303
Moca classes, Brazil Large: oblong screens 13/12 and 11; medium: 10; small: 9 and below Green coffee MAPA IN 82
Shell as defect, Brazil 3 shells = 1 defect Green coffee MAPA IN 82

The Mundo Novo text gives the lower bound as 5.34%, but the lowest value in its table is 5.94%; both are shown.9 In the 1955 study, the Portuguese summary names the fourth variety Bourbon Amarelo and the English summary Yellow Goiaba.5

Measuring the peaberry share#

ISO 4150 measures the peaberry share by mass. For a sample with many peaberries, a sieve with slotted openings is added to the nest of round-hole sieves.14 Slotted sieves are shaken parallel to the length of the slots, and beans stuck in a slot count as retained.14 Field studies use other bases: the Brazilian trial weighed the fraction on screens 10–13 in a 300 g sample,9 while the Colombian figure is a general statement without a stated basis.11 Shares from different studies are therefore not strictly comparable.

Do peaberries roast or cup differently?#

The idea is old. In Brittany, according to a coffee science textbook citing an older history of coffee, home roasters preferred peaberries because round beans rolled better in the pan and roasted more evenly.16 Measured studies are few, and they disagree:

  • Cenicafé, 2023. On the variety Cenicafé 1, bean shape affected every parameter tested. Peaberries dried faster, roasted more evenly with fewer scorched spots, broke under more uniform forces in grinding tests, and cupped as high as standard beans.11
  • Mexico, 2019. In three Catimor varieties, flat beans on screen 13 of the variety Colombia scored almost seven points higher than peaberries of the same screen.12 On screens 14 and 15, samples of mixed shape scored the same as or higher than flat beans, and the effect of shape depended on the variety.12

No study opened for this entry compared the chemical composition of peaberries and flat beans from the same lot. Roast behaviour by size and density is covered under bean size and density.

Open questions#

  • Share of each cause. Ovule abortion, chromosome damage, interspecific incompatibility and pollination failure are all reported routes to a peaberry. No study opened here measured how much each contributes in a commercial field.4,6,10
  • Climate and peaberries. Irrigation improved bean shape in Kenya, but no study opened here measured the effect of drought or heat on the peaberry share itself.8
  • Cup quality. The two cupping studies used different varieties and protocols and reached different results; a same-lot study across several varieties is missing.11,12

Key studies#

Santa Ram et al., 1990 – Peaberry development and breeding. Cut across fruits from the parents and three progenies of Selection 5 (Devamachy × S. 881) in India, 1984–1985. Peaberries took 18.54–31.18% of fruits and empty locules 5.35–29.74%, both higher in the progenies. The authors link peaberries to ovule abortion before fertilisation and empty locules to abnormalities after it, and propose selecting low-peaberry, self-compatible F3 plants.

Journal of Coffee Research 20(1), 69–76 · library record

Alves, 1955 – Polyembryony and empty locules in Coffea. Counted polyembryonic seeds, seeds without embryos and empty locules across arabica varieties and other species at Campinas. False polyembryony explained the elephant bean; empty locules reached about 30% in Mundo Novo and var. erecta.

Bragantia (Instituto Agronômico, Campinas), 1955 · SciELO

Duque Dussán et al., 2023 – Peaberry shape in post-harvest processing. Compared peaberries and standard beans of the variety Cenicafé 1 in drying, roasting, grinding and cupping. Shape affected each step: peaberries dried faster and roasted more evenly, and cupped as high as standard beans.

Journal of Food Process Engineering 46, e14461 · doi:10.1111/jfpe.14461

Luna González et al., 2019 – Bean size, shape and cup quality. Separated three Catimor varieties grown in Mexico by screen and by shape and cupped them. Flat beans beat peaberries on screen 13 in one variety, but mixed or small beans often cupped as well as larger flat beans.

International Journal of Food Properties 22, 758–767 · doi:10.1080/10942912.2019.1603997

Contested
  • Peaberries in the cup: in a 2019 Mexican study of three Catimor varieties, flat beans on screen 13 scored almost seven points higher than peaberries in the variety Colombia, and the effect of shape depended on variety12; a 2023 Cenicafé study on the variety Cenicafé 1 reports that peaberries reached the same high cup scores as standard beans.11
  • Whether peaberry production is inherited: a 1990 Indian breeding study proposes selecting individual plants with low peaberry production4, and irradiation raised the peaberry share through chromosomal aberrations6; a 2020 Ethiopian study concludes that peaberry may not be predominantly a heritable character.10
Share or cite this entry

Change log

  • – Removed the Ethiopian peaberry shares (3.24–11.07%, mean 7.31%, hybrid Gawe 16.22%) from the brief, quick facts, text, reference table and chart; brief range changed to about 5–10%. Reason: Owner ruling – Yilma and Kufa 2020 is Tier B and no independent Tier A/B source supports these numbers.
  • – Gene-expression study – four (not two) pectin methylesterase genes; removed “from the same plantings”. Reason: The paper names four PME genes and does not say the samples came from the same plantings.
  • – Indian breeding study (1990) – peaberry formation is now described as linked to, not shown to be caused by, the abortion of one ovule before fertilisation. Reason: The full text measures an association (regression of peaberries on free space in the ovary); the ovule abortion itself comes from earlier literature and unpublished cytology.

Sources

  1. BHerrera, J. C.; Lambot, C. (2017). The Coffee Tree – Genetic Diversity and Origin. In: Folmer, B. (ed.), The Craft and Science of Coffee, Academic Press, ch. 1, pp. 1–16. doi:10.1016/B978-0-12-803520-7.00001-3 · other copies
  2. AMinistério da Agricultura, Pecuária e Abastecimento (MAPA), Brazil (2003). Instrução Normativa nº 8, de 11 de junho de 2003 — Regulamento Técnico de Identidade e de Qualidade para a Classificação do Café Beneficiado Grão Cru. MAPA (copy hosted by Sindicafé-MG).
  3. AEast African Community (1999). EAS 130:1999 Green coffee beans – Specification. East African Community (copy hosted by Public.Resource.Org).
  4. ASanta Ram, A.; Sreenivasan, M. S.; Ramaiah, P. K. (1990). A study of peaberry development: its implications in coffee breeding. Journal of Coffee Research 20(1), 69–76 (Central Coffee Research Institute, India).
  5. AAlves, A. (1955). Melhoramento do cafeeiro. VIII – Novas determinações sôbre poliembrionia, sementes sem embrião e lojas dos frutos sem sementes em Coffea. Bragantia (Instituto Agronômico, Campinas).
  6. AMonge, S. F. (1962). Frecuencia de café caracolillo en plantas provenientes de semillas irradiadas. Turrialba 12(4) (IICA).
  7. ACarvalho, A.; Krug, C. A.; Mendes, J. E. T.; Antunes Filho, H.; Morais, H. de; Aloisi Sobrinho, J.; Morais, M. V. de; Rocha, T. R. da (1952). Melhoramento do cafeeiro: IV – Café Mundo Novo. Bragantia 12(4–6), 97–130. doi:10.1590/S0006-87051952000200001 · other copies
  8. AWormer, T. M. (1966). Shape of bean in Coffea arabica L. in Kenya. Turrialba 16(3) (IICA).
  9. ARaso, B. de S. M.; Botelho, C. E.; Rezende, J. C. de; Mendes, A. N. G.; Carvalho, G. R.; Ferreira, A. D. (2015). Avaliação agronômica de cafeeiros da cultivar Mundo Novo no Estado de Minas Gerais. Semina: Ciências Agrárias 36(4), 2363–2372.
  10. BYilma, A.; Kufa, T. (2020). Coffee Peaberry as A Potential Seed Source for Production. International Journal of Research Studies in Science, Engineering and Technology 7(9), 30–35.
  11. ADuque Dussán, E.; Figueroa Varela, P.; Sanz Uribe, J. R. (2023). Peaberry shape and size influence on different coffee postharvest processes. Journal of Food Process Engineering 46, e14461 (Spanish abstract on the Cenicafé publications page). doi:10.1111/jfpe.14461 · other copies
  12. ALuna González, A.; Macías López, A.; Taboada Gaytán, O. R.; Morales Ramos, V. (2019). Cup quality attributes of Catimors as affected by size and shape of coffee bean (Coffea arabica L.). International Journal of Food Properties 22, 758–767. doi:10.1080/10942912.2019.1603997 · other copies
  13. AFu, X.; Li, G.; Hu, F.; Huang, J.; Lou, Y.; Li, Y.; Li, Y.; He, H.; Lv, Y.; et al.; Cheng, J. (2023). Comparative transcriptome analysis in peaberry and regular bean coffee to identify bean quality associated genes. BMC Genomic Data 24, 12. doi:10.1186/s12863-022-01098-y · other copies
  14. AInternational Organization for Standardization (ISO/TC 34/SC 15) (2011). ISO 4150:2011 Green coffee or raw coffee — Size analysis — Manual and machine sieving. ISO.
  15. BInternational Trade Centre (ITC) (2011). The Coffee Exporter's Guide, Third Edition. International Trade Centre (UNCTAD/WTO), Geneva.
  16. BMorris, J. (2017). We Consumers – Tastes, Rituals, and Waves. In: Folmer, B. (ed.), The Craft and Science of Coffee, Academic Press, ch. 19. doi:10.1016/B978-0-12-803520-7.00019-0 · other copies
  17. BInternational Coffee Organization (2018). National Quality Standards (ICC 122-12). International Coffee Council, 122nd Session.

Tier A · authoritative · B · trade and expert · C · general. Supporting quotes for every source are in the data version of this entry. If a publisher’s site refuses access, try the doi link or “other copies”.

Suggest a change

On Peaberries and bean shapes. Tell us what you would change and why. A source that supports it helps most.