Wexflyer has already said. You are looking at a tab of paper sticking out the top attached to a later page - an file index marker, if scroll down several pages it is still there.
The page is annotated Marriage Register 1846 in Latin, previous entitled Baptisms for 1846 in English top right pane.
The contents of the microfilm & register are as indicated on the parish page and described on the cover.
https://registers.nli.ie/parishes/0553I've never seen the view that opens from your link without the buttons/ability to zoom, adjust contrast etc, or set full screen. Goes back to normal though when click on the
Microfilm 04620 / 01 makes reading easier. Seems to be view that opens in a new tab after clicking the Download button. From the standard View clicking on Glinsk takes you to the parish summary details (as link above) with date ranges and the microfilm numbers. The map on the right shows adjacent parishes which open when clicked on with their date ranges etc. Further west registers start later, some not till the 1850's or even 1860's.
When name searching I prefer to use Findmypast rather than Ancesty, is a free collection on Findmypast, all you need is a logon and the advanced search and image display options are I feel better. They split the collection into Baptisms, Marriages and Deaths in separate indexes.The transcriptions though are likely to be identical, unless error reports have been subsequently submitted, as Ancestry and Findmypast did it 50-50 and pooled the results.
https://search.findmypast.co.uk/search-world-records/ireland-roman-catholic-parish-marriagesThis is their entry for Hagerty-Leonard
https://findmypast.co.uk/transcript?id=IRE/PRS/MAR/1231949/1 which links to a full screen 'view original record' and to the NLI parish overview.
The microfilms are of varied age and quarlity as is the preservation of the registers and the handwriting of the ministers, scroll back and look at 1838-1839. There are often errors and omissions in the transcripts, and sometimes multiple pages are illegible.
Agree the register seems to have been batch updated with entries eg from a rough notebook/sheet into the book; hands do change. There may have been several chapels within the parish served by the same priest and curates, hence the variant forms of parish name:
Dunamon / Glinsk and Kilbegnet / Creggs / Kilbegnet / Killbegnet and Dunamon / Kilbegnet and Ballinakill / Killbegnet / Kilbegnet and Glinsk.
Some baptisms /marriages may even have been at home until 1850 see
https://www.rootschat.com/forum/index.php?topic=888326.msg7617128#msg7617128Glink is 80% in Co. Galway but 20% in Co. Roscommon.
Civil vs R.C parish map
https://www.johngrenham.com/places/rc_map_county_slider.php?county=GalwayThis explains some of the variants
https://glinsk.ie/history-of-the-church-in-glinsk/https://www.elphindiocese.ie/parishes-ministries/parish-details/?parishID=22Indiation: 1st line the groom & bride.
next line their witnesses, followed by the priests initials. [name may be on a previous page or from a list of clergy elsewhere]
Yes some of the witnesses then married being likely the same day (there are no dates for the 1st 2½ months of the year). On the right page, level with the Rafferty line 4 marriages have both lines indented - no consistancy.
If an entry is missing just manually create a citation. Add Source, select Ireland, Catholic Parish Registers, 1655-1915 from the drop down, put eg the parish name in the Detail box as that box needs something, and paste the Ancestry or NLI image URL in the web address box. I do similar for all Irish civil BMD from IrishGen or online sources that Ancestry does not have. Alternatively just save Mich Hegarty to your tree as the right image is connected (though you might then get hints for him from trees, and they for your Thos. Raftery).
How they were microfilmed is explained under
About at the top
https://registers.nli.ie/about but is more a case of how they were written and the size/quality of the book, which was down to the priest and parish and funds. Some are in Latin (Glinsk marriages are from 1855, always best enabling forname variants and not exact to allow for this), some perhaps few decades later pre-printed columns.
