P. 22 of Theorem List - Metamath Proof Explorer

Theorem List for Metamath Proof Explorer - 2101-2200   *Has distinct variable group(s)

Type	Label	Description
Statement
Theorem	pm11.07 2101	Axiom *11.07 in [WhiteheadRussell] p. 159. The original reads: *11.07 "Whatever possible argument 𝑥 may be, 𝜑(𝑥, 𝑦) is true whatever possible argument 𝑦 may be" implies the corresponding statement with 𝑥 and 𝑦 interchanged except in "𝜑(𝑥, 𝑦)". Under our formalism this appears to correspond to idi 1 and not to sbcom4 2100 as earlier thought. See https://groups.google.com/g/metamath/c/iS0fOvSemC8/m/M1zTH8wxCAAJ 2100. (Contributed by BJ, 16-Sep-2018.) (New usage is discouraged.)
⊢ 𝜑    ⇒   ⊢ 𝜑
Theorem	sbrimvw 2102*	Substitution in an implication with a variable not free in the antecedent affects only the consequent. Version of sbrim 2315 based on fewer axioms, but with more disjoint variable conditions. Based on an idea of GG. (Contributed by Wolf Lammen, 29-Jan-2024.)
⊢ ([𝑦 / 𝑥](𝜑 → 𝜓) ↔ (𝜑 → [𝑦 / 𝑥]𝜓))
Theorem	sbbiiev 2103*	An equivalence of substitutions (as in sbbii 2087) allowing the additional information that 𝑥 = 𝑡. Version of sbiev 2323 and sbievw 2104 without a disjoint variable condition on 𝜓, useful for substituting only part of 𝜑. (Contributed by SN, 24-Aug-2025.)
⊢ (𝑥 = 𝑡 → (𝜑 ↔ 𝜓))    ⇒   ⊢ ([𝑡 / 𝑥]𝜑 ↔ [𝑡 / 𝑥]𝜓)
Theorem	sbievw 2104*	Conversion of implicit substitution to explicit substitution. Version of sbie 2510 and sbiev 2323 with more disjoint variable conditions, requiring fewer axioms. (Contributed by NM, 30-Jun-1994.) (Revised by BJ, 18-Jul-2023.) (Proof shortened by SN, 24-Aug-2025.)
⊢ (𝑥 = 𝑦 → (𝜑 ↔ 𝜓))    ⇒   ⊢ ([𝑦 / 𝑥]𝜑 ↔ 𝜓)
Theorem	sbievwOLD 2105*	Obsolete version of sbievw 2104 as of 24-Aug-2025. (Contributed by NM, 30-Jun-1994.) (Revised by BJ, 18-Jul-2023.) (Proof modification is discouraged.) (New usage is discouraged.)
⊢ (𝑥 = 𝑦 → (𝜑 ↔ 𝜓))    ⇒   ⊢ ([𝑦 / 𝑥]𝜑 ↔ 𝜓)
Theorem	sbiedvw 2106*	Conversion of implicit substitution to explicit substitution (deduction version of sbievw 2104). Version of sbied 2511 and sbiedv 2512 with more disjoint variable conditions, requiring fewer axioms. (Contributed by NM, 30-Jun-1994.) (Revised by GG, 29-Jan-2024.)
⊢ ((𝜑 ∧ 𝑥 = 𝑦) → (𝜓 ↔ 𝜒))    ⇒   ⊢ (𝜑 → ([𝑦 / 𝑥]𝜓 ↔ 𝜒))
Theorem	2sbievw 2107*	Conversion of double implicit substitution to explicit substitution. Version of 2sbiev 2513 with more disjoint variable conditions, requiring fewer axioms. (Contributed by AV, 29-Jul-2023.) Avoid ax-13 2380. (Revised by GG, 10-Jan-2024.)
⊢ ((𝑥 = 𝑡 ∧ 𝑦 = 𝑢) → (𝜑 ↔ 𝜓))    ⇒   ⊢ ([𝑡 / 𝑥][𝑢 / 𝑦]𝜑 ↔ 𝜓)
Theorem	sbcom3vv 2108*	Substituting 𝑦 for 𝑥 and then 𝑧 for 𝑦 is equivalent to substituting 𝑧 for both 𝑥 and 𝑦. Version of sbcom3 2514 with a disjoint variable condition using fewer axioms. (Contributed by NM, 27-May-1997.) (Revised by Giovanni Mascellani, 8-Apr-2018.) (Revised by BJ, 30-Dec-2020.) (Proof shortened by Wolf Lammen, 19-Jan-2023.)
⊢ ([𝑧 / 𝑦][𝑦 / 𝑥]𝜑 ↔ [𝑧 / 𝑦][𝑧 / 𝑥]𝜑)
Theorem	sbievw2 2109*	sbievw 2104 applied twice, avoiding a DV condition on 𝑥, 𝑦. Based on proofs by Wolf Lammen. (Contributed by Steven Nguyen, 29-Jul-2023.)
⊢ (𝑥 = 𝑤 → (𝜑 ↔ 𝜒))    &   ⊢ (𝑤 = 𝑦 → (𝜒 ↔ 𝜓))    ⇒   ⊢ ([𝑦 / 𝑥]𝜑 ↔ 𝜓)
Theorem	sbco2vv 2110*	A composition law for substitution. Version of sbco2 2519 with disjoint variable conditions and fewer axioms. (Contributed by NM, 30-Jun-1994.) (Revised by BJ, 22-Dec-2020.) (Proof shortened by Wolf Lammen, 29-Apr-2023.)
⊢ ([𝑦 / 𝑧][𝑧 / 𝑥]𝜑 ↔ [𝑦 / 𝑥]𝜑)
Theorem	cbvsbv 2111*	Change the bound variable (i.e. the substituted one) in wff's linked by implicit substitution. The proof was extracted from a former cbvabv 2810 version. (Contributed by Wolf Lammen, 16-Mar-2025.)
⊢ (𝑥 = 𝑦 → (𝜑 ↔ 𝜓))    ⇒   ⊢ ([𝑧 / 𝑥]𝜑 ↔ [𝑧 / 𝑦]𝜓)
Theorem	sbco4lem 2112*	Lemma for sbco4 2113. It replaces the temporary variable 𝑣 with another temporary variable 𝑤. (Contributed by Jim Kingdon, 26-Sep-2018.) (Proof shortened by Wolf Lammen, 12-Oct-2024.) Avoid ax-11 2168. (Revised by SN, 3-Sep-2025.)
⊢ ([𝑥 / 𝑣][𝑦 / 𝑥][𝑣 / 𝑦]𝜑 ↔ [𝑥 / 𝑤][𝑦 / 𝑥][𝑤 / 𝑦]𝜑)
Theorem	sbco4 2113*	Two ways of exchanging two variables. Both sides of the biconditional exchange 𝑥 and 𝑦, either via two temporary variables 𝑢 and 𝑣, or a single temporary 𝑤. (Contributed by Jim Kingdon, 25-Sep-2018.) Avoid ax-11 2168. (Revised by SN, 3-Sep-2025.)
⊢ ([𝑦 / 𝑢][𝑥 / 𝑣][𝑢 / 𝑥][𝑣 / 𝑦]𝜑 ↔ [𝑥 / 𝑤][𝑦 / 𝑥][𝑤 / 𝑦]𝜑)
Theorem	equsb3 2114*	Substitution in an equality. (Contributed by Raph Levien and FL, 4-Dec-2005.) Reduce axiom usage. (Revised by Wolf Lammen, 23-Jul-2023.)
⊢ ([𝑦 / 𝑥]𝑥 = 𝑧 ↔ 𝑦 = 𝑧)
Theorem	equsb3r 2115*	Substitution applied to the atomic wff with equality. Variant of equsb3 2114. (Contributed by AV, 29-Jul-2023.) (Proof shortened by Wolf Lammen, 2-Sep-2023.)
⊢ ([𝑦 / 𝑥]𝑧 = 𝑥 ↔ 𝑧 = 𝑦)
Theorem	equsb1v 2116*	Substitution applied to an atomic wff. Version of equsb1 2499 with a disjoint variable condition, which neither requires ax-12 2189 nor ax-13 2380. (Contributed by NM, 10-May-1993.) (Revised by BJ, 11-Sep-2019.) Remove dependencies on axioms. (Revised by Wolf Lammen, 30-May-2023.) (Proof shortened by Steven Nguyen, 19-Jun-2023.) Revise df-sb 2074. (Revised by Steven Nguyen, 11-Jul-2023.) (Proof shortened by Steven Nguyen, 22-Jul-2023.)
⊢ [𝑦 / 𝑥]𝑥 = 𝑦
Theorem	nsb 2117	Any substitution in an always false formula is false. (Contributed by Steven Nguyen, 3-May-2023.)
⊢ (∀𝑥 ¬ 𝜑 → ¬ [𝑡 / 𝑥]𝜑)
Theorem	sbn1 2118	One direction of sbn 2291, using fewer axioms. Compare 19.2 1983. (Contributed by Steven Nguyen, 18-Aug-2023.)
⊢ ([𝑡 / 𝑥] ¬ 𝜑 → ¬ [𝑡 / 𝑥]𝜑)
1.4.9  Membership predicate
Syntax	wcel 2119	Extend wff definition to include the membership connective between classes. For a general discussion of the theory of classes, see mmset.html#class. The purpose of introducing wff 𝐴 ∈ 𝐵 here is to allow to prove the wel 2120 of predicate calculus in terms of the wcel 2119 of set theory, so that we do not overload the ∈ connective with two syntax definitions. This is done to prevent ambiguity that would complicate some Metamath parsers. The class variables 𝐴 and 𝐵 are introduced temporarily for the purpose of this definition but otherwise not used in predicate calculus. See df-clab 2719 for more information on the set theory usage of wcel 2119.
wff 𝐴 ∈ 𝐵
Theorem	wel 2120	Extend wff definition to include atomic formulas with the membership predicate. This is read either "𝑥 is an element of 𝑦", or "𝑥 is a member of 𝑦", or "𝑥 belongs to 𝑦", or "𝑦 contains 𝑥". Note: The phrase "𝑦 includes 𝑥 " means "𝑥 is a subset of 𝑦"; to use it also for 𝑥 ∈ 𝑦, as some authors occasionally do, is poor form and causes confusion, according to George Boolos (1992 lecture at MIT). This syntactic construction introduces a binary non-logical predicate symbol ∈ (stylized lowercase epsilon) into our predicate calculus. We will eventually use it for the membership predicate of set theory, but that is irrelevant at this point: the predicate calculus axioms for ∈ apply to any arbitrary binary predicate symbol. "Non-logical" means that the predicate is presumed to have additional properties beyond the realm of predicate calculus, although these additional properties are not specified by predicate calculus itself but rather by the axioms of a theory (in our case set theory) added to predicate calculus. "Binary" means that the predicate has two arguments. Instead of introducing wel 2120 as an axiomatic statement, as was done in an older version of this database, we introduce it by "proving" a special case of set theory's more general wcel 2119. This lets us avoid overloading the ∈ connective, thus preventing ambiguity that would complicate certain Metamath parsers. However, logically wel 2120 is considered to be a primitive syntax, even though here it is artificially "derived" from wcel 2119. Note: To see the proof steps of this syntax proof, type "MM> SHOW PROOF wel / ALL" in the Metamath program. (Contributed by NM, 24-Jan-2006.)
wff 𝑥 ∈ 𝑦
1.4.10  Axiom scheme ax-8 (Left Equality for Binary Predicate)
Axiom	ax-8 2121	Axiom of Left Equality for Binary Predicate. One of the equality and substitution axioms for a non-logical predicate in our predicate calculus with equality. It substitutes equal variables into the left-hand side of an arbitrary binary predicate ∈, which we will use for the set membership relation when set theory is introduced. This axiom scheme is a sub-scheme of Axiom Scheme B8 of system S2 of [Tarski], p. 75, whose general form cannot be represented with our notation. Also appears as Axiom scheme C12' in [Megill] p. 448 (p. 16 of the preprint). "Non-logical" means that the predicate is not a primitive of predicate calculus proper but instead is an extension to it. "Binary" means that the predicate has two arguments. In a system of predicate calculus with equality, like ours, equality is not usually considered to be a non-logical predicate. In systems of predicate calculus without equality, it typically would be. We prove in ax8 2125 that this axiom can be recovered from its weakened version ax8v 2122 where 𝑥 and 𝑦 are assumed to be disjoint variables. In particular, the only theorem referencing ax-8 2121 should be ax8v 2122. See the comment of ax8v 2122 for more details on these matters. (Contributed by NM, 30-Jun-1993.) (Revised by BJ, 7-Dec-2020.) Use ax8 2125 instead. (New usage is discouraged.)
⊢ (𝑥 = 𝑦 → (𝑥 ∈ 𝑧 → 𝑦 ∈ 𝑧))
Theorem	ax8v 2122*	Weakened version of ax-8 2121, with a disjoint variable condition on 𝑥, 𝑦. This should be the only proof referencing ax-8 2121, and it should be referenced only by its two weakened versions ax8v1 2123 and ax8v2 2124, from which ax-8 2121 is then rederived as ax8 2125, which shows that either ax8v 2122 or the conjunction of ax8v1 2123 and ax8v2 2124 is sufficient. (Contributed by BJ, 7-Dec-2020.) Use ax8 2125 instead. (New usage is discouraged.)
⊢ (𝑥 = 𝑦 → (𝑥 ∈ 𝑧 → 𝑦 ∈ 𝑧))
Theorem	ax8v1 2123*	First of two weakened versions of ax8v 2122, with an extra disjoint variable condition on 𝑥, 𝑧, see comments there. (Contributed by BJ, 7-Dec-2020.)
⊢ (𝑥 = 𝑦 → (𝑥 ∈ 𝑧 → 𝑦 ∈ 𝑧))
Theorem	ax8v2 2124*	Second of two weakened versions of ax8v 2122, with an extra disjoint variable condition on 𝑦, 𝑧 see comments there. (Contributed by BJ, 7-Dec-2020.)
⊢ (𝑥 = 𝑦 → (𝑥 ∈ 𝑧 → 𝑦 ∈ 𝑧))
Theorem	ax8 2125	Proof of ax-8 2121 from ax8v1 2123 and ax8v2 2124, proving sufficiency of the conjunction of the latter two weakened versions of ax8v 2122, which is itself a weakened version of ax-8 2121. (Contributed by BJ, 7-Dec-2020.) (Proof shortened by Wolf Lammen, 11-Apr-2021.)
⊢ (𝑥 = 𝑦 → (𝑥 ∈ 𝑧 → 𝑦 ∈ 𝑧))
Theorem	elequ1 2126	An identity law for the non-logical predicate. (Contributed by NM, 30-Jun-1993.)
⊢ (𝑥 = 𝑦 → (𝑥 ∈ 𝑧 ↔ 𝑦 ∈ 𝑧))
Theorem	elsb1 2127*	Substitution for the first argument of the non-logical predicate in an atomic formula. See elsb2 2136 for substitution for the second argument. (Contributed by NM, 7-Nov-2006.) (Proof shortened by Andrew Salmon, 14-Jun-2011.) Reduce axiom usage. (Revised by Wolf Lammen, 24-Jul-2023.)
⊢ ([𝑦 / 𝑥]𝑥 ∈ 𝑧 ↔ 𝑦 ∈ 𝑧)
Theorem	cleljust 2128*	When the class variables in Definition df-clel 2815 are replaced with setvar variables, this theorem of predicate calculus is the result. This theorem provides part of the justification for the consistency of that definition, which "overloads" the setvar variables in wel 2120 with the class variables in wcel 2119. (Contributed by NM, 28-Jan-2004.) Revised to use equsexvw 2012 in order to remove dependencies on ax-10 2152, ax-12 2189, ax-13 2380. Note that there is no disjoint variable condition on 𝑥, 𝑦, that is, on the variables of the left-hand side, as should be the case for definitions. (Revised by BJ, 29-Dec-2020.)
⊢ (𝑥 ∈ 𝑦 ↔ ∃𝑧(𝑧 = 𝑥 ∧ 𝑧 ∈ 𝑦))
1.4.11  Axiom scheme ax-9 (Right Equality for Binary Predicate)
Axiom	ax-9 2129	Axiom of Right Equality for Binary Predicate. One of the equality and substitution axioms for a non-logical predicate in our predicate calculus with equality. It substitutes equal variables into the right-hand side of an arbitrary binary predicate ∈, which we will use for the set membership relation when set theory is introduced. This axiom scheme is a sub-scheme of Axiom Scheme B8 of system S2 of [Tarski], p. 75, whose general form cannot be represented with our notation. Also appears as Axiom scheme C13' in [Megill] p. 448 (p. 16 of the preprint). We prove in ax9 2133 that this axiom can be recovered from its weakened version ax9v 2130 where 𝑥 and 𝑦 are assumed to be disjoint variables. In particular, the only theorem referencing ax-9 2129 should be ax9v 2130. See the comment of ax9v 2130 for more details on these matters. (Contributed by NM, 21-Jun-1993.) (Revised by BJ, 7-Dec-2020.) Use ax9 2133 instead. (New usage is discouraged.)
⊢ (𝑥 = 𝑦 → (𝑧 ∈ 𝑥 → 𝑧 ∈ 𝑦))
Theorem	ax9v 2130*	Weakened version of ax-9 2129, with a disjoint variable condition on 𝑥, 𝑦. This should be the only proof referencing ax-9 2129, and it should be referenced only by its two weakened versions ax9v1 2131 and ax9v2 2132, from which ax-9 2129 is then rederived as ax9 2133, which shows that either ax9v 2130 or the conjunction of ax9v1 2131 and ax9v2 2132 is sufficient. (Contributed by BJ, 7-Dec-2020.) Use ax9 2133 instead. (New usage is discouraged.)
⊢ (𝑥 = 𝑦 → (𝑧 ∈ 𝑥 → 𝑧 ∈ 𝑦))
Theorem	ax9v1 2131*	First of two weakened versions of ax9v 2130, with an extra disjoint variable condition on 𝑥, 𝑧, see comments there. (Contributed by BJ, 7-Dec-2020.)
⊢ (𝑥 = 𝑦 → (𝑧 ∈ 𝑥 → 𝑧 ∈ 𝑦))
Theorem	ax9v2 2132*	Second of two weakened versions of ax9v 2130, with an extra disjoint variable condition on 𝑦, 𝑧 see comments there. (Contributed by BJ, 7-Dec-2020.)
⊢ (𝑥 = 𝑦 → (𝑧 ∈ 𝑥 → 𝑧 ∈ 𝑦))
Theorem	ax9 2133	Proof of ax-9 2129 from ax9v1 2131 and ax9v2 2132, proving sufficiency of the conjunction of the latter two weakened versions of ax9v 2130, which is itself a weakened version of ax-9 2129. (Contributed by BJ, 7-Dec-2020.) (Proof shortened by Wolf Lammen, 11-Apr-2021.)
⊢ (𝑥 = 𝑦 → (𝑧 ∈ 𝑥 → 𝑧 ∈ 𝑦))
Theorem	elequ2 2134	An identity law for the non-logical predicate. (Contributed by NM, 21-Jun-1993.)
⊢ (𝑥 = 𝑦 → (𝑧 ∈ 𝑥 ↔ 𝑧 ∈ 𝑦))
Theorem	elequ2g 2135*	A form of elequ2 2134 with a universal quantifier. Its converse is the axiom of extensionality ax-ext 2712. (Contributed by BJ, 3-Oct-2019.)
⊢ (𝑥 = 𝑦 → ∀𝑧(𝑧 ∈ 𝑥 ↔ 𝑧 ∈ 𝑦))
Theorem	elsb2 2136*	Substitution for the second argument of the non-logical predicate in an atomic formula. See elsb1 2127 for substitution for the first argument. (Contributed by Rodolfo Medina, 3-Apr-2010.) (Proof shortened by Andrew Salmon, 14-Jun-2011.) Reduce axiom usage. (Revised by Wolf Lammen, 24-Jul-2023.)
⊢ ([𝑦 / 𝑥]𝑧 ∈ 𝑥 ↔ 𝑧 ∈ 𝑦)
Theorem	elequ12 2137	An identity law for the non-logical predicate, which combines elequ1 2126 and elequ2 2134. The analogous theorems for class terms are eleq1 2828, eleq2 2829, and eleq12 2830 respectively. (Contributed by BJ, 29-Sep-2019.)
⊢ ((𝑥 = 𝑦 ∧ 𝑧 = 𝑡) → (𝑥 ∈ 𝑧 ↔ 𝑦 ∈ 𝑡))
Theorem	ru0 2138*	The FOL statement used in the standard proof of Russell's paradox ru 3728. (Contributed by NM, 7-Aug-1994.) Extract from proof of ru 3728 and reduce axiom usage. (Revised by BJ, 12-Oct-2019.)
⊢ ¬ ∀𝑥(𝑥 ∈ 𝑦 ↔ ¬ 𝑥 ∈ 𝑥)
1.4.12  Logical redundancy of ax-10 , ax-11 , ax-12 , ax-13 The original axiom schemes of Tarski's predicate calculus are ax-4 1816, ax-5 1917, ax6v 1975, ax-7 2015, ax-8 2121, and ax-9 2129, together with rule ax-gen 1802. See mmset.html#compare 1802. They are given as axiom schemes B4 through B8 in [KalishMontague] p. 81. These are shown to be logically complete by Theorem 1 of [KalishMontague] p. 85. The axiom system of set.mm includes the auxiliary axiom schemes ax-10 2152, ax-11 2168, ax-12 2189, and ax-13 2380, which are not part of Tarski's axiom schemes. Each object-language instance of them is provable from Tarski's axioms, so they are logically redundant. However, they are conjectured not to be provable directly as schemes from Tarski's axiom schemes using only Metamath's direct substitution rule. They are used to make our system "scheme complete", i.e., able to prove directly all possible schemes with wff and setvar variables, bundled or not, whose object-language instances are valid. (ax-12 2189 has been proved to be required; see https://us.metamath.org/award2003.html#9a 2189. Metalogical independence of the other three are open problems.) (There are additional predicate calculus axiom schemes included in set.mm such as ax-c5 39376, but they can all be proved as theorems from the above.) Terminology: Two setvar (individual) metavariables are "bundled" in an axiom or theorem scheme when there is no distinct variable constraint ($d) imposed on them. (The term "bundled" is due to Raph Levien.) For example, the 𝑥 and 𝑦 in ax-6 1974 are bundled, but they are not in ax6v 1975. We also say that a scheme is bundled when it has at least one pair of bundled setvar variables. If distinct variable conditions are added to all setvar variable pairs in a bundled scheme, we call that the "principal" instance of the bundled scheme. For example, ax6v 1975 is the principal instance of ax-6 1974. Whenever a common variable is substituted for two or more bundled variables in an axiom or theorem scheme, we call the substitution instance "degenerate". For example, the instance ¬ ∀𝑥¬ 𝑥 = 𝑥 of ax-6 1974 is degenerate. An advantage of bundling is ease of use since there are fewer distinct variable restrictions ($d) to be concerned with, and theorems are more general. There may be some economy in being able to prove facts about principal and degenerate instances simultaneously. A disadvantage is that bundling may present difficulties in translations to other proof languages, which typically lack the concept (in part because their variables often represent the variables of the object language rather than metavariables ranging over them). Because Tarski's axiom schemes are logically complete, they can be used to prove any object-language instance of ax-10 2152, ax-11 2168, ax-12 2189, and ax-13 2380. "Translating" this to Metamath, it means that Tarski's axioms can prove any substitution instance of ax-10 2152, ax-11 2168, ax-12 2189, or ax-13 2380 in which (1) there are no wff metavariables and (2) all setvar variables are mutually distinct i.e. are not bundled. In effect this is mimicking the object language by pretending that each setvar variable is an object-language variable. (There may also be specific instances with wff metavariables and/or bundling that are directly provable from Tarski's axiom schemes, but it isn't guaranteed. Whether all of them are possible is part of the still open metalogical independence problem for our additional axiom schemes.) It can be useful to see how this can be done, both to show that our additional schemes are valid metatheorems of Tarski's system and to be able to translate object-language instances of our proofs into proofs that would work with a system using only Tarski's original schemes. In addition, it may (or may not) provide insight into the conjectured metalogical independence of our additional schemes. The theorem schemes ax10w 2140, ax11w 2141, ax12w 2144, and ax13w 2147 are derived using only Tarski's axiom schemes, showing that Tarski's schemes can be used to derive all substitution instances of ax-10 2152, ax-11 2168, ax-12 2189, and ax-13 2380 meeting Conditions (1) and (2). (The "w" suffix stands for "weak version".) Each hypothesis of ax10w 2140, ax11w 2141, and ax12w 2144 is of the form (𝑥 = 𝑦 → (𝜑 ↔ 𝜓)) where 𝜓 is an auxiliary or "dummy" wff metavariable in which 𝑥 doesn't occur. We can show by induction on formula length that the hypotheses can be eliminated in all cases meeting Conditions (1) and (2). The example ax12wdemo 2146 illustrates the techniques (equality theorems and bound variable renaming) used to achieve this. We also show the degenerate instances for axioms with bundled variables in ax11dgen 2142, ax12dgen 2145, ax13dgen1 2148, ax13dgen2 2149, ax13dgen3 2150, and ax13dgen4 2151. (Their proofs are trivial, but we include them to be thorough.) Combining the principal and degenerate cases outside of Metamath, we show that the bundled schemes ax-10 2152, ax-11 2168, ax-12 2189, and ax-13 2380 are schemes of Tarski's system, meaning that all object-language instances they generate are theorems of Tarski's system. It is interesting that Tarski used the bundled scheme ax-6 1974 in an older system, so it seems the main purpose of his later ax6v 1975 was just to show that the weaker unbundled form is sufficient rather than an aesthetic objection to bundled free and bound variables. Since we adopt the bundled ax-6 1974 as our official axiom, we show that the degenerate instance holds in ax6dgen 2139. (Recall that in set.mm, the only statement referencing ax-6 1974 is ax6v 1975.) The case of sp 2195 is curious: originally an axiom scheme of Tarski's system, it was proved logically redundant by Lemma 9 of [KalishMontague] p. 86. However, the proof is by induction on formula length, and the scheme form ∀𝑥𝜑 → 𝜑 apparently cannot be proved directly from Tarski's other axiom schemes. The best we can do seems to be spw 2041, again requiring substitution instances of 𝜑 that meet Conditions (1) and (2) above. Note that our direct proof sp 2195 requires ax-12 2189, which is not part of Tarski's system.
Theorem	ax6dgen 2139	Tarski's system uses the weaker ax6v 1975 instead of the bundled ax-6 1974, so here we show that the degenerate case of ax-6 1974 can be derived. Even though ax-6 1974 is in the list of axioms used, recall that in set.mm, the only statement referencing ax-6 1974 is ax6v 1975. We later rederive from ax6v 1975 the bundled form as ax6 2392 with the help of the auxiliary axiom schemes. (Contributed by NM, 23-Apr-2017.)
⊢ ¬ ∀𝑥 ¬ 𝑥 = 𝑥
Theorem	ax10w 2140*	Weak version of ax-10 2152 from which we can prove any ax-10 2152 instance not involving wff variables or bundling. Uses only Tarski's FOL axiom schemes. It is an alias of hbn1w 2055 introduced for labeling consistency. (Contributed by NM, 9-Apr-2017.) Use hbn1w 2055 instead. (New usage is discouraged.)
⊢ (𝑥 = 𝑦 → (𝜑 ↔ 𝜓))    ⇒   ⊢ (¬ ∀𝑥𝜑 → ∀𝑥 ¬ ∀𝑥𝜑)
Theorem	ax11w 2141*	Weak version of ax-11 2168 from which we can prove any ax-11 2168 instance not involving wff variables or bundling. Uses only Tarski's FOL axiom schemes. Unlike ax-11 2168, this theorem requires that 𝑥 and 𝑦 be distinct i.e. are not bundled. It is an alias of alcomimw 2050 introduced for labeling consistency. (Contributed by NM, 10-Apr-2017.) Use alcomimw 2050 instead. (New usage is discouraged.)
⊢ (𝑦 = 𝑧 → (𝜑 ↔ 𝜓))    ⇒   ⊢ (∀𝑥∀𝑦𝜑 → ∀𝑦∀𝑥𝜑)
Theorem	ax11dgen 2142	Degenerate instance of ax-11 2168 where bundled variables 𝑥 and 𝑦 have a common substitution. Uses only Tarski's FOL axiom schemes. (Contributed by NM, 13-Apr-2017.)
⊢ (∀𝑥∀𝑥𝜑 → ∀𝑥∀𝑥𝜑)
Theorem	ax12wlem 2143*	Lemma for weak version of ax-12 2189. Uses only Tarski's FOL axiom schemes. In some cases, this lemma may lead to shorter proofs than ax12w 2144. (Contributed by NM, 10-Apr-2017.)
⊢ (𝑥 = 𝑦 → (𝜑 ↔ 𝜓))    ⇒   ⊢ (𝑥 = 𝑦 → (𝜑 → ∀𝑥(𝑥 = 𝑦 → 𝜑)))
Theorem	ax12w 2144*	Weak version of ax-12 2189 from which we can prove any ax-12 2189 instance not involving wff variables or bundling. Uses only Tarski's FOL axiom schemes. An instance of the first hypothesis will normally require that 𝑥 and 𝑦 be distinct (unless 𝑥 does not occur in 𝜑). For an example of how the hypotheses can be eliminated when we substitute an expression without wff variables for 𝜑, see ax12wdemo 2146. (Contributed by NM, 10-Apr-2017.)
⊢ (𝑥 = 𝑦 → (𝜑 ↔ 𝜓))    &   ⊢ (𝑦 = 𝑧 → (𝜑 ↔ 𝜒))    ⇒   ⊢ (𝑥 = 𝑦 → (∀𝑦𝜑 → ∀𝑥(𝑥 = 𝑦 → 𝜑)))
Theorem	ax12dgen 2145	Degenerate instance of ax-12 2189 where bundled variables 𝑥 and 𝑦 have a common substitution. Uses only Tarski's FOL axiom schemes. (Contributed by NM, 13-Apr-2017.)
⊢ (𝑥 = 𝑥 → (∀𝑥𝜑 → ∀𝑥(𝑥 = 𝑥 → 𝜑)))
Theorem	ax12wdemo 2146*	Example of an application of ax12w 2144 that results in an instance of ax-12 2189 for a contrived formula with mixed free and bound variables, (𝑥 ∈ 𝑦 ∧ ∀𝑥𝑧 ∈ 𝑥 ∧ ∀𝑦∀𝑧𝑦 ∈ 𝑥), in place of 𝜑. The proof illustrates bound variable renaming with cbvalvw 2043 to obtain fresh variables to avoid distinct variable clashes. Uses only Tarski's FOL axiom schemes. (Contributed by NM, 14-Apr-2017.)
⊢ (𝑥 = 𝑦 → (∀𝑦(𝑥 ∈ 𝑦 ∧ ∀𝑥 𝑧 ∈ 𝑥 ∧ ∀𝑦∀𝑧 𝑦 ∈ 𝑥) → ∀𝑥(𝑥 = 𝑦 → (𝑥 ∈ 𝑦 ∧ ∀𝑥 𝑧 ∈ 𝑥 ∧ ∀𝑦∀𝑧 𝑦 ∈ 𝑥))))
Theorem	ax13w 2147*	Weak version (principal instance) of ax-13 2380. (Because 𝑦 and 𝑧 don't need to be distinct, this actually bundles the principal instance and the degenerate instance (¬ 𝑥 = 𝑦 → (𝑦 = 𝑦 → ∀𝑥𝑦 = 𝑦)).) Uses only Tarski's FOL axiom schemes. The proof is trivial but is included to complete the set ax10w 2140, ax11w 2141, and ax12w 2144. (Contributed by NM, 10-Apr-2017.)
⊢ (¬ 𝑥 = 𝑦 → (𝑦 = 𝑧 → ∀𝑥 𝑦 = 𝑧))
Theorem	ax13dgen1 2148	Degenerate instance of ax-13 2380 where bundled variables 𝑥 and 𝑦 have a common substitution. Uses only Tarski's FOL axiom schemes. (Contributed by NM, 13-Apr-2017.)
⊢ (¬ 𝑥 = 𝑥 → (𝑥 = 𝑧 → ∀𝑥 𝑥 = 𝑧))
Theorem	ax13dgen2 2149	Degenerate instance of ax-13 2380 where bundled variables 𝑥 and 𝑧 have a common substitution. Uses only Tarski's FOL axiom schemes. (Contributed by NM, 13-Apr-2017.)
⊢ (¬ 𝑥 = 𝑦 → (𝑦 = 𝑥 → ∀𝑥 𝑦 = 𝑥))
Theorem	ax13dgen3 2150	Degenerate instance of ax-13 2380 where bundled variables 𝑦 and 𝑧 have a common substitution. Uses only Tarski's FOL axiom schemes. (Contributed by NM, 13-Apr-2017.)
⊢ (¬ 𝑥 = 𝑦 → (𝑦 = 𝑦 → ∀𝑥 𝑦 = 𝑦))
Theorem	ax13dgen4 2151	Degenerate instance of ax-13 2380 where bundled variables 𝑥, 𝑦, and 𝑧 have a common substitution. Therefore, also a degenerate instance of ax13dgen1 2148, ax13dgen2 2149, and ax13dgen3 2150. Also an instance of the intuitionistic tautology pm2.21 123. Uses only Tarski's FOL axiom schemes. (Contributed by NM, 13-Apr-2017.) Reduce axiom usage. (Revised by Wolf Lammen, 10-Oct-2021.)
⊢ (¬ 𝑥 = 𝑥 → (𝑥 = 𝑥 → ∀𝑥 𝑥 = 𝑥))
1.5  Predicate calculus with equality: Auxiliary axiom schemes (4 schemes) In this section we introduce four additional schemes ax-10 2152, ax-11 2168, ax-12 2189, and ax-13 2380 that are not part of Tarski's system but can be proved (outside of Metamath) as theorem schemes of Tarski's system. These are needed to give our system the property of "scheme completeness", which means that we can prove (with Metamath) all possible theorem schemes expressible in our language of wff metavariables ranging over object-language wffs, and setvar variables ranging over object-language individual variables. To show that these schemes are valid metatheorems of Tarski's system S2, above we proved from Tarski's system theorems ax10w 2140, ax11w 2141, ax12w 2144, and ax13w 2147, which show that any object-language instance of these schemes (emulated by having no wff metavariables and requiring all setvar variables to be mutually distinct) can be proved using only the schemes in Tarski's system S2. An open problem is to show that these four additional schemes are mutually metalogically independent and metalogically independent from Tarski's. So far, independence of ax-12 2189 from all others has been shown, and independence of Tarski's ax-6 1974 from all others has been shown; see items 9a and 11 on https://us.metamath.org/award2003.html 1974.
1.5.1  Axiom scheme ax-10 (Quantified Negation)
Axiom	ax-10 2152	Axiom of Quantified Negation. Axiom C5-2 of [Monk2] p. 113. This axiom scheme is logically redundant (see ax10w 2140) but is used as an auxiliary axiom scheme to achieve scheme completeness. It means that 𝑥 is not free in ¬ ∀𝑥𝜑. (Contributed by NM, 21-May-2008.) Use its alias hbn1 2153 instead if you must use it. Any theorem in first-order logic (FOL) that contains only set variables that are all mutually distinct, and has no wff variables, can be proved *without* using ax-10 2152 through ax-13 2380, by invoking ax10w 2140 through ax13w 2147. We encourage proving theorems *without* ax-10 2152 through ax-13 2380 and moving them up to the ax-4 1816 through ax-9 2129 section. (New usage is discouraged.)
⊢ (¬ ∀𝑥𝜑 → ∀𝑥 ¬ ∀𝑥𝜑)
Theorem	hbn1 2153	Alias for ax-10 2152 to be used instead of it. (Contributed by NM, 24-Jan-1993.) (Proof shortened by Wolf Lammen, 18-Aug-2014.)
⊢ (¬ ∀𝑥𝜑 → ∀𝑥 ¬ ∀𝑥𝜑)
Theorem	hbe1 2154	The setvar 𝑥 is not free in ∃𝑥𝜑. Corresponds to the axiom (5) of modal logic (see also modal5 2166). (Contributed by NM, 24-Jan-1993.)
⊢ (∃𝑥𝜑 → ∀𝑥∃𝑥𝜑)
Theorem	hbe1a 2155	Dual statement of hbe1 2154. Modified version of axc7e 2327 with a universally quantified consequent. (Contributed by Wolf Lammen, 15-Sep-2021.)
⊢ (∃𝑥∀𝑥𝜑 → ∀𝑥𝜑)
Theorem	nf5-1 2156	One direction of nf5 2293 can be proved with a smaller footprint on axiom usage. (Contributed by Wolf Lammen, 16-Sep-2021.)
⊢ (∀𝑥(𝜑 → ∀𝑥𝜑) → Ⅎ𝑥𝜑)
Theorem	nf5i 2157	Deduce that 𝑥 is not free in 𝜑 from the definition. (Contributed by Mario Carneiro, 11-Aug-2016.)
⊢ (𝜑 → ∀𝑥𝜑)    ⇒   ⊢ Ⅎ𝑥𝜑
Theorem	nf5dh 2158	Deduce that 𝑥 is not free in 𝜓 in a context. (Contributed by Mario Carneiro, 24-Sep-2016.) df-nf 1791 changed. (Revised by Wolf Lammen, 11-Oct-2021.)
⊢ (𝜑 → ∀𝑥𝜑)    &   ⊢ (𝜑 → (𝜓 → ∀𝑥𝜓))    ⇒   ⊢ (𝜑 → Ⅎ𝑥𝜓)
Theorem	nf5dv 2159*	Apply the definition of not-free in a context. (Contributed by Mario Carneiro, 11-Aug-2016.) df-nf 1791 changed. (Revised by Wolf Lammen, 18-Sep-2021.) (Proof shortened by Wolf Lammen, 13-Jul-2022.)
⊢ (𝜑 → (𝜓 → ∀𝑥𝜓))    ⇒   ⊢ (𝜑 → Ⅎ𝑥𝜓)
Theorem	nfnaew 2160*	All variables are effectively bound in a distinct variable specifier. Version of nfnae 2442 with a disjoint variable condition, which does not require ax-13 2380. (Contributed by Mario Carneiro, 11-Aug-2016.) Avoid ax-13 2380. (Revised by GG, 10-Jan-2024.) (Proof shortened by Wolf Lammen, 25-Sep-2024.)
⊢ Ⅎ𝑧 ¬ ∀𝑥 𝑥 = 𝑦
Theorem	nfe1 2161	The setvar 𝑥 is not free in ∃𝑥𝜑. (Contributed by Mario Carneiro, 11-Aug-2016.)
⊢ Ⅎ𝑥∃𝑥𝜑
Theorem	nfa1 2162	The setvar 𝑥 is not free in ∀𝑥𝜑. (Contributed by Mario Carneiro, 11-Aug-2016.) df-nf 1791 changed. (Revised by Wolf Lammen, 11-Sep-2021.) Remove dependency on ax-12 2189. (Revised by Wolf Lammen, 12-Oct-2021.)
⊢ Ⅎ𝑥∀𝑥𝜑
Theorem	nfna1 2163	A convenience theorem particularly designed to remove dependencies on ax-11 2168 in conjunction with distinctors. (Contributed by Wolf Lammen, 2-Sep-2018.)
⊢ Ⅎ𝑥 ¬ ∀𝑥𝜑
Theorem	nfia1 2164	Lemma 23 of [Monk2] p. 114. (Contributed by Mario Carneiro, 24-Sep-2016.)
⊢ Ⅎ𝑥(∀𝑥𝜑 → ∀𝑥𝜓)
Theorem	nfnf1 2165	The setvar 𝑥 is not free in Ⅎ𝑥𝜑. (Contributed by Mario Carneiro, 11-Aug-2016.) Remove dependency on ax-12 2189. (Revised by Wolf Lammen, 12-Oct-2021.)
⊢ Ⅎ𝑥Ⅎ𝑥𝜑
Theorem	modal5 2166	The analogue in our predicate calculus of axiom (5) of modal logic S5. See also hbe1 2154. (Contributed by NM, 5-Oct-2005.)
⊢ (¬ ∀𝑥 ¬ 𝜑 → ∀𝑥 ¬ ∀𝑥 ¬ 𝜑)
Theorem	nfs1v 2167*	The setvar 𝑥 is not free in [𝑦 / 𝑥]𝜑 when 𝑥 and 𝑦 are distinct. (Contributed by Mario Carneiro, 11-Aug-2016.) Shorten nfs1v 2167 and hbs1 2285 combined. (Revised by Wolf Lammen, 28-Jul-2022.)
⊢ Ⅎ𝑥[𝑦 / 𝑥]𝜑
1.5.2  Axiom scheme ax-11 (Quantifier Commutation)
Axiom	ax-11 2168	Axiom of Quantifier Commutation. This axiom says universal quantifiers can be swapped. Axiom scheme C6' in [Megill] p. 448 (p. 16 of the preprint). Also appears as Lemma 12 of [Monk2] p. 109 and Axiom C5-3 of [Monk2] p. 113. This axiom scheme is logically redundant (see ax11w 2141) but is used as an auxiliary axiom scheme to achieve metalogical completeness. Use its weak version alcomimw 2050 when it allows to avoid dependence on ax-11 2168. (Contributed by NM, 12-Mar-1993.)
⊢ (∀𝑥∀𝑦𝜑 → ∀𝑦∀𝑥𝜑)
Theorem	alcoms 2169	Swap quantifiers in an antecedent. (Contributed by NM, 11-May-1993.)
⊢ (∀𝑥∀𝑦𝜑 → 𝜓)    ⇒   ⊢ (∀𝑦∀𝑥𝜑 → 𝜓)
Theorem	alcom 2170	Theorem 19.5 of [Margaris] p. 89. Use its weak version alcomw 2052 when it allows to avoid dependence on ax-11 2168. (Contributed by NM, 30-Jun-1993.)
⊢ (∀𝑥∀𝑦𝜑 ↔ ∀𝑦∀𝑥𝜑)
Theorem	alrot3 2171	Theorem *11.21 in [WhiteheadRussell] p. 160. (Contributed by Andrew Salmon, 24-May-2011.)
⊢ (∀𝑥∀𝑦∀𝑧𝜑 ↔ ∀𝑦∀𝑧∀𝑥𝜑)
Theorem	alrot4 2172	Rotate four universal quantifiers twice. (Contributed by NM, 2-Feb-2005.) (Proof shortened by Fan Zheng, 6-Jun-2016.)
⊢ (∀𝑥∀𝑦∀𝑧∀𝑤𝜑 ↔ ∀𝑧∀𝑤∀𝑥∀𝑦𝜑)
Theorem	excom 2173	Theorem 19.11 of [Margaris] p. 89. (Contributed by NM, 5-Aug-1993.) Remove dependencies on ax-5 1917, ax-6 1974, ax-7 2015, ax-10 2152, ax-12 2189. (Revised by Wolf Lammen, 8-Jan-2018.) (Proof shortened by Wolf Lammen, 22-Aug-2020.)
⊢ (∃𝑥∃𝑦𝜑 ↔ ∃𝑦∃𝑥𝜑)
Theorem	excomim 2174	One direction of Theorem 19.11 of [Margaris] p. 89. (Contributed by NM, 5-Aug-1993.) (Revised by Mario Carneiro, 24-Sep-2016.) Remove dependencies on ax-5 1917, ax-6 1974, ax-7 2015, ax-10 2152, ax-12 2189. (Revised by Wolf Lammen, 8-Jan-2018.)
⊢ (∃𝑥∃𝑦𝜑 → ∃𝑦∃𝑥𝜑)
Theorem	excom13 2175	Swap 1st and 3rd existential quantifiers. (Contributed by NM, 9-Mar-1995.)
⊢ (∃𝑥∃𝑦∃𝑧𝜑 ↔ ∃𝑧∃𝑦∃𝑥𝜑)
Theorem	exrot3 2176	Rotate existential quantifiers. (Contributed by NM, 17-Mar-1995.)
⊢ (∃𝑥∃𝑦∃𝑧𝜑 ↔ ∃𝑦∃𝑧∃𝑥𝜑)
Theorem	exrot4 2177	Rotate existential quantifiers twice. (Contributed by NM, 9-Mar-1995.)
⊢ (∃𝑥∃𝑦∃𝑧∃𝑤𝜑 ↔ ∃𝑧∃𝑤∃𝑥∃𝑦𝜑)
Theorem	hbal 2178	If 𝑥 is not free in 𝜑, it is not free in ∀𝑦𝜑. (Contributed by NM, 12-Mar-1993.)
⊢ (𝜑 → ∀𝑥𝜑)    ⇒   ⊢ (∀𝑦𝜑 → ∀𝑥∀𝑦𝜑)
Theorem	hbald 2179	Deduction form of bound-variable hypothesis builder hbal 2178. (Contributed by NM, 2-Jan-2002.)
⊢ (𝜑 → ∀𝑦𝜑)    &   ⊢ (𝜑 → (𝜓 → ∀𝑥𝜓))    ⇒   ⊢ (𝜑 → (∀𝑦𝜓 → ∀𝑥∀𝑦𝜓))
Theorem	sbal 2180*	Move universal quantifier in and out of substitution. (Contributed by NM, 16-May-1993.) (Proof shortened by Wolf Lammen, 29-Sep-2018.) Reduce dependencies on axioms. (Revised by Steven Nguyen, 13-Aug-2023.)
⊢ ([𝑧 / 𝑦]∀𝑥𝜑 ↔ ∀𝑥[𝑧 / 𝑦]𝜑)
Theorem	sbalv 2181*	Quantify with new variable inside substitution. (Contributed by NM, 18-Aug-1993.)
⊢ ([𝑦 / 𝑥]𝜑 ↔ 𝜓)    ⇒   ⊢ ([𝑦 / 𝑥]∀𝑧𝜑 ↔ ∀𝑧𝜓)
Theorem	hbsbw 2182*	If 𝑧 is not free in 𝜑, it is not free in [𝑦 / 𝑥]𝜑 when 𝑦 and 𝑧 are distinct. Version of hbsb 2532 with a disjoint variable condition, which requires fewer axioms. (Contributed by NM, 12-Aug-1993.) Remove dependencies on axioms. (Revised by GG, 23-May-2024.) (Proof shortened by Wolf Lammen, 14-May-2025.)
⊢ (𝜑 → ∀𝑧𝜑)    ⇒   ⊢ ([𝑦 / 𝑥]𝜑 → ∀𝑧[𝑦 / 𝑥]𝜑)
Theorem	sbcom2 2183*	Commutativity law for substitution. Used in proof of Theorem 9.7 of [Megill] p. 449 (p. 16 of the preprint). (Contributed by NM, 27-May-1997.) (Proof shortened by Wolf Lammen, 23-Dec-2022.)
⊢ ([𝑤 / 𝑧][𝑦 / 𝑥]𝜑 ↔ [𝑦 / 𝑥][𝑤 / 𝑧]𝜑)
Theorem	sbco4lemOLD 2184*	Obsolete version of sbco4lem 2112 as of 3-Sep-2025. (Contributed by Jim Kingdon, 26-Sep-2018.) (Proof shortened by Wolf Lammen, 12-Oct-2024.) (Proof modification is discouraged.) (New usage is discouraged.)
⊢ ([𝑥 / 𝑣][𝑦 / 𝑥][𝑣 / 𝑦]𝜑 ↔ [𝑥 / 𝑤][𝑦 / 𝑥][𝑤 / 𝑦]𝜑)
Theorem	sbco4OLD 2185*	Obsolete version of sbco4 2113 as of 3-Sep-2025. (Contributed by Jim Kingdon, 25-Sep-2018.) (Proof modification is discouraged.) (New usage is discouraged.)
⊢ ([𝑦 / 𝑢][𝑥 / 𝑣][𝑢 / 𝑥][𝑣 / 𝑦]𝜑 ↔ [𝑥 / 𝑤][𝑦 / 𝑥][𝑤 / 𝑦]𝜑)
Theorem	nfa2 2186	Lemma 24 of [Monk2] p. 114. (Contributed by Mario Carneiro, 24-Sep-2016.) Remove dependency on ax-12 2189. (Revised by Wolf Lammen, 18-Oct-2021.)
⊢ Ⅎ𝑥∀𝑦∀𝑥𝜑
Theorem	nfexhe 2187	Version of nfex 2333 with the existential dual to the 'h' hypothesis, avoiding ax-12 2189. (Contributed by SN, 11-Feb-2026.)
⊢ (∃𝑥𝜑 → 𝜑)    ⇒   ⊢ Ⅎ𝑥∃𝑦𝜑
Theorem	nfexa2 2188	An inner universal quantifier's variable is bound. (Contributed by SN, 11-Feb-2026.)
⊢ Ⅎ𝑥∃𝑦∀𝑥𝜑
1.5.3  Axiom scheme ax-12 (Substitution)
Axiom	ax-12 2189	Axiom of Substitution. One of the 5 equality axioms of predicate calculus. The final consequent ∀𝑥(𝑥 = 𝑦 → 𝜑) is a way of expressing "𝑦 substituted for 𝑥 in wff 𝜑 " (cf. sb6 2096). It is based on Lemma 16 of [Tarski] p. 70 and Axiom C8 of [Monk2] p. 105, from which it can be proved by cases. The original version of this axiom was ax-c15 39382 and was replaced with this shorter ax-12 2189 in Jan. 2007. The old axiom is proved from this one as Theorem axc15 2430. Conversely, this axiom is proved from ax-c15 39382 as Theorem ax12 2431. Juha Arpiainen proved the metalogical independence of this axiom (in the form of the older axiom ax-c15 39382) from the others on 19-Jan-2006. See item 9a at https://us.metamath.org/award2003.html 39382. See ax12v 2190 and ax12v2 2191 for other equivalents of this axiom that (unlike this axiom) have distinct variable restrictions. This axiom scheme is logically redundant (see ax12w 2144) but is used as an auxiliary axiom scheme to achieve scheme completeness. (Contributed by NM, 22-Jan-2007.) (New usage is discouraged.)
⊢ (𝑥 = 𝑦 → (∀𝑦𝜑 → ∀𝑥(𝑥 = 𝑦 → 𝜑)))
Theorem	ax12v 2190*	This is essentially Axiom ax-12 2189 weakened by additional restrictions on variables. Besides axc11r 2376, this theorem should be the only one referencing ax-12 2189 directly. Both restrictions on variables have their own value. If for a moment we assume 𝑥 could be set to 𝑦, then, after elimination of the tautology 𝑦 = 𝑦, immediately we have 𝜑 → ∀𝑦𝜑 for all 𝜑 and 𝑦, that is ax-5 1917, a degenerate result. The second restriction is not necessary, but a simplification that makes the following interpretation easier to see. Since 𝜑 textually at most depends on 𝑥, we can look at it at some given 'fixed' 𝑦. This theorem now states that the truth value of 𝜑 will stay constant, as long as we 'vary 𝑥 around 𝑦' only such that 𝑥 = 𝑦 still holds. Or in other words, equality is the finest grained logical expression. If you cannot differ two sets by =, you won't find a whatever sophisticated expression that does. One might wonder how the described variation of 𝑥 is possible at all. Note that Metamath is a text processor that easily sees a difference between text chunks {𝑥 ∣ ¬ 𝑥 = 𝑥} and {𝑦 ∣ ¬ 𝑦 = 𝑦}. Our usual interpretation is to abstract from textual variations of the same set, but we are free to interpret Metamath's formalism differently, and in fact let 𝑥 run through all textual representations of sets. Had we allowed 𝜑 to depend also on 𝑦, this idea is both harder to see, and it is less clear that this extra freedom introduces effects not covered by other axioms. (Contributed by Wolf Lammen, 8-Aug-2020.)
⊢ (𝑥 = 𝑦 → (𝜑 → ∀𝑥(𝑥 = 𝑦 → 𝜑)))
Theorem	ax12v2 2191*	It is possible to remove any restriction on 𝜑 in ax12v 2190. Same as Axiom C8 of [Monk2] p. 105. Use ax12v 2190 instead when sufficient. (Contributed by NM, 5-Aug-1993.) Remove dependencies on ax-10 2152 and ax-13 2380. (Revised by Jim Kingdon, 15-Dec-2017.) (Proof shortened by Wolf Lammen, 8-Dec-2019.)
⊢ (𝑥 = 𝑦 → (𝜑 → ∀𝑥(𝑥 = 𝑦 → 𝜑)))
Theorem	ax12ev2 2192*	Version of ax12v2 2191 rewritten to use an existential quantifier. One direction of sbalex 2254 without the universal quantifier, avoiding ax-10 2152. (Contributed by SN, 14-Aug-2025.)
⊢ (∃𝑥(𝑥 = 𝑦 ∧ 𝜑) → (𝑥 = 𝑦 → 𝜑))
Theorem	19.8a 2193	If a wff is true, it is true for at least one instance. Special case of Theorem 19.8 of [Margaris] p. 89. See 19.8v 1990 for a version with a disjoint variable condition requiring fewer axioms. (Contributed by NM, 9-Jan-1993.) Allow a shortening of sp 2195. (Revised by Wolf Lammen, 13-Jan-2018.) (Proof shortened by Wolf Lammen, 8-Dec-2019.)
⊢ (𝜑 → ∃𝑥𝜑)
Theorem	19.8ad 2194	If a wff is true, it is true for at least one instance. Deduction form of 19.8a 2193. (Contributed by DAW, 13-Feb-2017.)
⊢ (𝜑 → 𝜓)    ⇒   ⊢ (𝜑 → ∃𝑥𝜓)
Theorem	sp 2195	Specialization. A universally quantified wff implies the wff without a quantifier. Axiom scheme B5 of [Tarski] p. 67 (under his system S2, defined in the last paragraph on p. 77). Also appears as Axiom scheme C5' in [Megill] p. 448 (p. 16 of the preprint). This corresponds to the axiom (T) of modal logic. For the axiom of specialization presented in many logic textbooks, see Theorem stdpc4 2079. This theorem shows that our obsolete axiom ax-c5 39376 can be derived from the others. The proof uses ideas from the proof of Lemma 21 of [Monk2] p. 114. It appears that this scheme cannot be derived directly from Tarski's axioms without auxiliary axiom scheme ax-12 2189. It is thought the best we can do using only Tarski's axioms is spw 2041. Also see spvw 1988 where 𝑥 and 𝜑 are disjoint, using fewer axioms. (Contributed by NM, 21-May-2008.) (Proof shortened by Scott Fenton, 24-Jan-2011.) (Proof shortened by Wolf Lammen, 13-Jan-2018.)
⊢ (∀𝑥𝜑 → 𝜑)
Theorem	spi 2196	Inference rule of universal instantiation, or universal specialization. Converse of the inference rule of (universal) generalization ax-gen 1802. Contrary to the rule of generalization, its closed form is valid, see sp 2195. (Contributed by NM, 5-Aug-1993.)
⊢ ∀𝑥𝜑    ⇒   ⊢ 𝜑
Theorem	sps 2197	Generalization of antecedent. (Contributed by NM, 5-Jan-1993.)
⊢ (𝜑 → 𝜓)    ⇒   ⊢ (∀𝑥𝜑 → 𝜓)
Theorem	2sp 2198	A double specialization (see sp 2195). Another double specialization, closer to PM*11.1, is 2stdpc4 2081. (Contributed by BJ, 15-Sep-2018.)
⊢ (∀𝑥∀𝑦𝜑 → 𝜑)
Theorem	spsd 2199	Deduction generalizing antecedent. (Contributed by NM, 17-Aug-1994.)
⊢ (𝜑 → (𝜓 → 𝜒))    ⇒   ⊢ (𝜑 → (∀𝑥𝜓 → 𝜒))
Theorem	19.2g 2200	Theorem 19.2 of [Margaris] p. 89, generalized to use two setvar variables. Use 19.2 1983 when sufficient. (Contributed by Mel L. O'Cat, 31-Mar-2008.)
⊢ (∀𝑥𝜑 → ∃𝑦𝜑)