Applying each marriage law's own way of counting kinship to every kind of cousin out to ninth cousins, we found that Taiwan's Civil Code reaches four kinds, second cousins among them. The Catholic Church's 1983 Code of Canon Law reaches one, first cousins, and makes that marriage invalid unless a dispensation is given. Singapore's Women's Charter and Malaysia's Law Reform (Marriage and Divorce) Act, the civil marriage statutes for non-Muslims, name no cousin. Each kind is fixed by the generations from each person up to their shared ancestor. The degree is one less than the smaller count, and the difference between the counts is how many times removed.
Each removal spans one generation between you and the cousin
Share a grandparent and you are first cousins, two generations down on each side. Share a great-grandparent and you are second cousins, three and three. Your parent's first cousin is two generations below your great-grandparent and you are three, so the pair is first cousins once removed. Your first cousin's child is the same in reverse.
Second cousin and first cousin once removed are each one step beyond a first cousin, in different directions. A second cousin is in your own generation. A first cousin once removed is a generation older or younger, and your shared ancestor is a grandparent to one of you and a great-grandparent to the other. The term reads the same from either end.
Canon law's degree count now spans both lines, as Taiwan's does
Taiwan's Civil Code §968 counts up to the common ancestor, then down to the relative, and takes 總世數, the total number of generations, as the degree. First cousins are four degrees apart. Canon law has agreed since 1983: under can. 108 §3 there are "as many degrees as there are persons in both the lines together", leaving out the common ancestor.
Can. 96 §3 of the 1917 Code counted only the longer line, so first cousins were in the second degree. Its can. 1076 §2 made marriage invalid up to the third degree, which took in three kinds of cousin: first cousins, first cousins once removed and second cousins. The 1983 Code took effect on the first Sunday of Advent that year, 27 November by our calendar arithmetic. Its can. 1091 §2 makes marriage invalid "up to and including the fourth degree", first cousins only. Can. 1078 §3 rules out a dispensation in the second degree, brother and sister, so a first-cousin impediment can be dispensed.
Taiwan's Civil Code stretches its bar to second cousins
Section 983 says collateral blood relatives 在六親等以內者, within the sixth degree, may not marry. The one exception is for same-generation relatives of the fourth and sixth degree related through adoption. Our count of the four kinds of cousin it reaches:
| Relationship | Generations to the shared ancestor | Degree, Taiwan and 1983 canon law | Degree, 1917 canon law |
|---|---|---|---|
| First cousin | 2 and 2 | 4 | 2 |
| First cousin once removed | 2 and 3 | 5 | 3 |
| First cousin twice removed | 2 and 4 | 6 | 4 |
| Second cousin | 3 and 3 | 6 | 3 |
Mainland China's Civil Code, art. 1048, reads 「直系血亲或者三代以内的旁系血亲禁止结婚。」: lineal relatives, or collateral relatives within three generations, may not marry. The code does not define how a 代 is counted, and we found no interpretation from the National People's Congress. The civil-affairs authorities of Shenzhen (December 2024) and Guangdong province (December 2023) read it as covering people descended from the same grandparents, and the Shenzhen page gives 表兄妹, first cousins, as an example.
Singapore's and Malaysia's statute lists extend to no cousin
Under s.10(1) of Singapore's Women's Charter 1961, a marriage to anyone listed in the First Schedule "is void", though under s.10(2) the Minister may license one valid under the custom or religion that would have applied. The schedule's 46 entries run from parent and grandparent to father's sister and sister's son, with no cousin and no great-aunt. Section 3(4) excludes marriages between Muslims.
Malaysia's Act 164, as reprinted to 1 January 2006, goes a generation further. Section 11(1) names the great-aunt, great-uncle, great-niece and great-nephew, and no cousin, and s.11(6) lets the Chief Minister license a marriage it would otherwise bar. The 2017 amending Act, A1546, left s.11 unchanged. Section 3(3) says the Act does not apply to a Muslim. Singapore and Malaysia are the only ASEAN members whose statutes we checked, and we did not research their rules for Muslim marriages.
Chinese 堂 extends only through an unbroken male line
Taiwan's Ministry of Education dictionary defines 堂兄弟 as the sons of one's father's brothers, and 表兄弟 as the sons of a father's sister or a mother's brother or sister. So father → brother → son is 堂, and father → sister → son is 表, though both are on the father's side. Both entries cover first cousins only. Fuller term lists are in our guides to what to call relatives in Chinese and to Malay and Indonesian family terms.
Until 7 October 2026 our cousin relationship calculator took 堂 or 表 from a dropdown that defaulted to 表, whatever path had been built, so the page's own example, father → brother → son, gave 表兄弟姐妹. Across generation counts up to ten it also misnamed 14 of the 18 aunt-and-niece pairs: wherever the gap was three generations or more, it called the younger side a grand-niece or grand-nephew, so a great-great-aunt's great-grand-niece became a grand-niece. Every cousin name in the same sweep was right. The fix of 7 October reads the line from the path for first cousins, labels the same rule an extension for other cousins, and names great-grand-nieces. The live page may show the old version until it is deployed.
The spread around the expected share stretches with each generation out
For cousins descended from an ancestral couple, the expected share of DNA is (1/2)^(a + b - 1), where a and b are each person's generations to those ancestors. That gives 12.5% for first cousins and 3.125% for second cousins, matching all five cousin values in Hill and Weir's 2011 model, a theoretical genome rather than a sample of people. It also puts a half-uncle and a great-grandparent at the first-cousin 12.5%, the kind of overlap our grandparent and half-sibling DNA guide covers.
The model's standard deviation, as a share of the expected value, roughly doubles with each cousin degree: 17% for first cousins and 38% for second. For third cousins it reaches 71% (our arithmetic). Measuring real relatives, Caballero and colleagues (2019) found the middle half of first cousins sharing between 11.1% and 13.8%, after the authors shifted their data's mean to the theoretical value.
Shared segments extend to fewer cousins at each further degree
The same paper simulated 10,000 pairs for each relationship. Every simulated first and second cousin pair shared at least one segment. Among third cousins, 1.5% shared no segment of any size. The share with none rose to 27.3% for fourth cousins and 67.4% for fifth. By sixth cousins it was 88.9%, so in that simulation barely one pair in nine still carried any piece of the ancestors their family tree says they share.
