Let $X\subset \mathbb P^n$ be a smooth hypersurface over base field $\mathbb C$. A pencil of hyperplanes is just a projective line $(X_t)$ in $\mathbb P^{n*}$. It is called a Lefschetz pencil if it satisfies the following: (I followed the definition in SGA)

(i) The axis (intersection of $X_t$'s) intersects with $X$ transversally.

(ii) Almost all (enough for one) $X_t$ intersects with $X$ transversally.

(iii) The non-transversal intersection in (ii) contains one node.

And on SGA it also proves that almost all pencils are Lefschetz. Now I want to know how many are those non-Lefschetz pencils, more precisely:

Does there exist some $X$, on which all the non-Lefschetz pencils cover the $\mathbb P^{n*}$? i.e. for any hyperplane section $H$, we can find non-Lefschetz pencil $(X_t)$ pass through it.

I guess it does not exist, but have no idea how to prove it. Any hints or references would be helpful.


I should remove the condition (i) otherwise it is trivial.


Assume there is a single hyperplane section $X_0$ of $X$ which has worse singularity than just a node. Then any pencil containing $X_0$ is non-Lefschetz, and these pencils cover all the $\mathbb{P}^{n*}$.

Of course, if there is no such $X_0$ as above, this construction does not work. However, I think the only case when such $X_0$ does not exist is that of $X$ being a smooth quadric. In that case the projective dual variety to $X$ is yet another smooth quadric, and lines in $\mathbb{P}^{n*}$ tangent to it provide non-Lefschetz pencils that cover all the $\mathbb{P}^{n*}$.

| cite | improve this answer | |
  • $\begingroup$ Thanks for your answer. Btw can I ask another question (maybe I should start a new question but anyway) why we need the condition (i) in the definition? Isn’t it followed by (ii)? $\endgroup$ – User X Nov 17 '18 at 9:39
  • $\begingroup$ Please, give a precise reference to the definition. $\endgroup$ – Sasha Nov 17 '18 at 10:12
  • $\begingroup$ I followed the definition in Groupes de monodromie en géométrie algébrique. II. Lecture Notes in Mathematics, Vol. 340. Séminaire de Géométrie Algébrique du Bois-Marie 1967–1969 (SGA 7 II). $\endgroup$ – User X Nov 17 '18 at 11:20
  • $\begingroup$ A precise reference should include expose and section numbers. In fact "axis" ("l'axe") in English is usually called "base locus". It surely doesn't follow from (ii). For instance, take $X \subset \mathbb{P}^2$ to be a smooth conic, and the pencil of lines through a point ON $X$. $\endgroup$ – Sasha Nov 17 '18 at 14:44
  • $\begingroup$ Sorry I thought something wrong. Thanks. $\endgroup$ – User X Nov 17 '18 at 19:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.