"Nadia talent" schreef
I still believe (see below) that this probably needs to be corrected to
var. longissima André, only a check of the publication can confirm.

***
I disagree about the "probable". It would not hurt to check the original
publication by André (there is a 1983 translation available), but this is
presumably superfluous. If you don't want to believe Mez (I do) then it is
would be easiest to check a recent monograph on Tillandsia.
* * *

Tillandsia usneoides var. longissima (André) Mez, 1896
I still believe (see below) that this is probably a later homonym.

ICBN Art 35.4. If in one whole publication (Art 35.5), prior to 1
January 1890, only one infraspecific rank is admitted, it is considered to
be that of variety unless this would be contrary to the author's statements
in the same publication."

presumably does not apply as Mez (in the Flora Brasiliensis) indicates
that André published this name in the rank of forma. If this is correct then
we have an "author's statement" and there is no need to involve Art 35.4.

I disagree. The publication in question is André's.

***
The relevant publication is André's, but I see no reason to disbelieve Mez
when he reports on what is there. Authors like him were painstakingly
accurate in such minutiae. Besides, TROPICOS agrees with Mez.
* * *

However, it would be advisable
to check André's publication to make sure that he did not use varieties
also, or state that he considered the formas to be subsidiary to some other
infraspecific rank used by other authors.

***
I am sorry, but that is not relevant. Art 35.4 is designed for those cases
where the author does not indicate which rank he is using (as in the
/Species plantarum/, where infraspecific taxa are marked by Greek letters
alpha, beta, gamma: these names are taken to be varieties). If André did
indicate that he regarded this as a forma, that is the end of it: thereby it
is a forma. It does not matter if he used varietes as well or not.
* * *

As you note, the GCI and Mez disagree about the rank used by André, but
in the absence of evidence to the contrary I will assume that the CGI is in
error (this is a fairly safe assumption as indexes are notorious for the
amount of error they contain). This is supported by TROPICOS which states
that Mez did transfer the name from the rank of forma to that of varietas
(upgrading this plant in rank), although not in the Flora Brasiliensis but
in the Monographiae Phanerogamarum

Interesting! I had not realized that TROPICOS and IPNI are not
synchronized.

***
TROPICOS and IPNI have very little in common. Their basic nature is entirely
different (do see the FAQ of this NG). IPNI is a compilation of three
different nomenclatural indexes: it just registers what names were
published. IPNI is full of errors (don't get me started). TROPICOS is run by
the Missouri Botanical Garden and is a taxonomic database so it contains
flora's and monographs (hard taxonomic data). There are some errors in
TROPICOS, but you really need to know what you are doing to find them.
* * *

Monographiae Phanerogamum is unfortunately not in Gallica or any other
online database that I know of.

So IPNI appears to need just the one record to be added (but the original
documents need to be checked, as always):

"Tillandsia usneoides f. longissima Mez Fl. Bras. (Martius) 3, pt. 3: 615.
1894."

***
Possible but not very likely. See above.
* * *

There would be no need to add:

"Tillandsia usneoides var. longissima Mez Monogr. Phan. 9: 883. 1896."
because it is just a later homonym of André's var. (but it would be nice if
IPNI listed all homonyms and clearly labelled them as such, as well as all
corrections that have been made including those that were made long ago in
the GCI and IK).

TROPICOS apparently needs to have the formae listed from André changed to
vars. to keep it up to date with the Codes of Nomenclature.

As to the classification IPNI is a nomenclatural database, which means
they are listing botanical names, and a botanical name never has more than
three parts (so Tillandsia usneoides var conspecta f. longissima is not a
name, but a classification).

But please do not let me discourage you!
PvR

I fear that we are discouraging others from attempting to resolve
nomenclatural problems. They aren't always this difficult, but it is
necessary to accumulate all the old literature, and it can take years to
accomplish that for a large group of plants. I strongly believe that we all
have a responsibility to submit corrections to the database editors as we
discover them while delving into that old literature. It helps, of course,
if one doesn't work on too many plants at once.

***
As nomenclatural problems go this is straightforward. As to
correcting the databases, actually it does help to deal with a lot of
literature, this will increase familarity with such issues, which will help
in knowing what to report and how. The editors in IPNI most likely will go
back to the original literature themselves (so as to avoid introducing new
errors), and it helps to indicate the potential problems as accurately as
possible.
PvR