Hypothetical reconstruction of Late Roman Helmet evolution, based on new research that's been coming out over the past decade and a half or so. (1) 
Going from left to right, the first to look at are ridge helmets. Commonly divided into just two types, there are actually 6 distinct variants of ridge helmets. The first of these is the "Comb" Helmet,, typified by the examples from Berkasovo-I and Budapest.
Next are what I classify as "Quadripartite-Type Ridge Helmets." Formerly usually referred to as the "Berkasovo-Type" which I found inadequate because there were fundamentally different constructions for many of these helmets. Only two have been found: Burgh Castle and Concesti.
These helmets are the most similar to the Berkasovo-I Comb Helmet. They have a four-piece (Quadripartite) bowl, a base ring around the bottom, large cheekpieces usually with a secondary rectangular plate, and a nasal. The difference is that these are proper ridge helmets.
The next in the line are the "Bipartite A-Type Ridge Helmet." These are characterized by having a two-piece bowl conjoined by a central ridge, a base ring, large cheekpieces usually with a secondary plate (or plates), and a nasal. These are typified by Berkasovo-II, Heteny, etc.
From these comes, probably, the "Bipartite B-Type Ridge Helmet." These are characterized by having a two-piece bowl conjoined by a central ridge, no base ring, large cheekpieces, and an optional nasal. This is typified by the 5th century helmet found at Iatrus-Krivina.
The fourth type is the "Bipartite C-Type Ridge Helmet." Formerly called the "Intercisa-Type." These have a bipartite bowl with "attic-style" cheekpieces, usually no nasal, and no base ring. With some slight exceptions. These are characterized by the Intercisa-Dunjavaros helmets.
The last ridge helmet type is the Bieberweir-Type Imitation Ridge Helmet, which isn't a ridge helmet at all. It has a one piece bowl, an outer base ring, a nasal, a maille neckguard, and large cheekpieces (the only surviving example of which are asiatic-style).
Interestingly, two of these have been found, so it's debatable whether these were Germanic or Roman.
Next in the line(s) are the spangenhelmets and bandhelmets. These are an interesting topic because they might appear as early as on Trajan's Column (below). Although I think the artist may have had some access to captured Dacian helmets, they're still heavily stylized.
I think it's possible that some of these helmets were late Bosphoran-style Chalcidean Helmets, like those shown below, which may have lasted beyond the 1st century BC. It's possible the artist may have been depicting helmets of this style due to his classical training.
Archaeologically speaking, we have to separate the evolution of Bandhelmets and Spangenhelmets. Spangenhelmets have a weird evolution that involves several different styles of helmet. But Bandhelmets are much simpler and I think originated as imitations of Roman or Greek helmets.
Nothing remains of the Bretzenheim helmet but a plaster mould. It is very arbitrarily dated from the 1st-5th centuries AD.
However, typologically this helmet might not date before the late 5th century, as it is most similar to the Trivieres and Shorwell helmets.
However, typologically this helmet might not date before the late 5th century, as it is most similar to the Trivieres and Shorwell helmets.
That makes the earliest known Bandhelmet a Hunnic find from Turaeovo, dating between 400 and 450 AD. The riveting style may suggest a Sassanid manufacture, but I think it was probably a Hun original. But the widespread distribution of the bandhelmets says they're older than this.
Roman Bandhelmets are typified by the examples from Narona, Sveti Vid, and several other locations. They can be identified by the characteristic three rivets in a triangular pattern, indicating their Roman (or sometimes Germanic) manufacture. These date to the early 6th century.
Kreuzbandhelmets break off from this, which are typified by Trivieres, Shorwell, and probably Bretzenheim as well. The examples from Groningen, dating to the 7th century, have clear Roman influence with silver bands or internal base rings. Also it's spiky and I love it.
The last of those, the Groningen Bandhelmet, bears similarity with the direction Roman Bandhelmets took. In the late 7th century we have a find in the Brooklyn Museum that shows they clearly merged with Intercisa-Type Helmets. This probably occurred in the late 5th century.
Roman Bandhelmets also crossed with Baldenheims in the late 5th Century as those developed, as evidenced by a single find from Voivoda, Bulgaria.
Bandhelmets would eventually develop into multi-piece constructions during the middle ages, such as these 10th century examples from the Netherlands and Qairawan. They are commonly depicted in Byzantine Art (3rd image shows such a reconstruction of a Byzantine variant).
But Voivoda brings us to the next line of helmets: The Baldenheims. These were a uniquely Roman style of helmet that apparently replaced the fancier variants of ridge helmets. They have a six-piece bowl conjoined by metal bands and a base ring that forms "teardrop" shaped panels.
You can see the clear Steppe/Sassanian influence: the "feathering" on the cheekpiece and the cheekpiece's shape are both eastern styles. The question of their origin remains open, but it's clear they share a common ancestor with the Sassanian Mesopotamian-Type Helmets (shown):
That being said the Romano-Germanic Kreuzbandhelmets like Shorwell are probably related to the Sassanid Mesopotamian-Type Helmets too.
But we have no evidence for what preceeded these three styles. The Mesopotamian-Type, Baldenheim-Type, and Kreuzbandhelmet all suddenly appear in the late 5th-early 6th centuries. The earliest one is Stuttgart, which dates to 470 AD. There's nothing before then.
Back from Lunch! Baldenheims do somehow come out of Spangenhelmets. We know that much. The problem is we don't know too much about the typology of Spangenhelmets either.
Some of the helmets on Trajan's column bear a marked similarity to those found in Khutor City, dating to the 1st and 2nd centuries AD and being of Sarmatian origin. This is where it gets interesting, because the Khutor City helmets are kind of a mutt type of helmet.
As a result we start getting into questions as to the origins of Lamellar Helmets, Segment Helmets, Spangenhelmets, and Skeletal Helmets all at once. Because the Russian Steppe finds are kind of a mess to organize. Here's a chart of the 3rd-5th century finds.
I'll stay on course with the diagram above and tackle the Tarasovo Grave 6 and Grave 1784 Helmets first (top left and bottom left, respectively). Tarasovo Grave 1784 (shown) is interesting because it appears to show a progression towards the Leiden-Type Spangenhelmet.
Tarasovo 1784 isn't alone, in fact. It has a sister that was found in Russia, due to be published in the coming years. The helmets have a riveting pattern in the Sassanid Style: the closely-packed continuous riveting up the conjoining bands and two around the rim of the bowl.
The logical path of spread must be a Sassanid origin that was transmitted into the Balkans via Steppe Nomad expansion in the 5th century. These developed into the earliest datable Roman spangenhelmets, the Novae-Kopiae Helmets from the Iron Gates fortress in Bulgaria.
Going back to Tarasovo, these appear to show a progressive evolution from the Grave 1784 and the Grave 6 helmets. A new conical cap on the tip of the bowl now conjoins the bands with four rivets, instead of a single large brass rivet. A base ring has been added. Reconstruction:
The debate over the Novae/Leiden-Type Spangenhelmets, though comes from the Arch of Galerius. Roman soldiers can be seen wearing some sort of conical segmented helmet, and that's where the debate arises.
Again we have to break the Galerius arch down further. So we'll use Pavel Simak's wonderful detail painting here as a guide. What's immediately noticeable is that we can see two types of helmet: one with many bands of construction, and single-piece conical helmets.
Adam Kubik has suggested these are lamellar helmets of the Kispek-Type. This bears some issues: 1. Kispek is Alan. 2. Kispek is not a Lamellar Helmet, it's a hybrid composite of a Skeletal and Lamellar Helmet, as Christian Miks shows.
I don't have a good image but the top left of Dr. Miks' slide in his conference shows the composite frame of the Kispek Helmet. And the lack of a conical cap and cheekpieces are clearly indicative that these are also not the helmets on Galerius' arch.
(Top right, sorry). Anyways, Miks relates the Galerius Arch to Kalkhni-type Spangenhelmets. The Helmet from Kalkhni (bottom right) is a Hunnic example from Dagestan dating to the 5th century. It doesn't seem to have followed the same evolutionary pathway as the Kispek Helmet.
Although Miks rules out the Kalkhni for the Deir-El-Medinah on Galerius' arch (coming up), this first hypothesis is probably correct. The Kalkhni-Type Lamellar Helmets probably did not have a base ring with Roman-style ears. It probably had cheekpieces. See the similarity?
(Drawing above by Seb Herzynia, who is making a replica). Anyways, the Kalkhni Helmet derives from a tradition of Lamellar Helmets that comes out of Central Asia, rather than Kispek which is a Pontic style. The earliest Pontic/Volga finds are from the 2nd century AD.
Kispek comes from the Stanica Tibilisskaya Helmet, probably, which is dated to the 2nd Century. Kispek is basically this, but with more plates (the original also probably had a wood, not a leather bowl. Although leather over wood is possible, like shield construction).
And from the Pontic Lamellar helmets we not only get the Arch of Galerius helmets, but also the later Avar style. And the Roman ones too, like this one found in a Roman arms manufacturing center at Stara Zagora, dating to about 580.
But now back to the Spangenhelmets. So where do they fit in? The Deir El Medinah and Leiden Museum Spangenhelmets, the former from Deir El Medinah Egypt and the latter *maybe* from Memphis, are really arbitrarily dated as 3rd-7th centuries.
Clearly they're not the Galerius arch helmets. Their construction is very similar, and Deir El Medinah is probably a Romanized modification. The only similar finds which have been dated are Novae, dated to the 580's, and Jerusalem, dated to 614-634.
Deir el Medinah is a rather romanized helmet and I would hesitantly put it as possibly Justinianic, because of its six-piece construction like most Baldenheims, while the later 6th and 7th century helmets move to four piece construction (including Baldenheims).
E.g. Herakleia Lynkestis, a Roman Baldenheim originally dated to 480 but re-dated to the Avar destruction of the site, being IIRC also 614 AD.
This doesn't mean Spangenhelmets weren't in use earlier than the 6th century by the Roman army. The Dura Europus Graffito clearly shows a Khutor City type helmet.
Although I would hesitantly classify these instead as Segment Helmets, despite the Khutor City ones being weird hybrids. This 2nd Century AD example from Itsyatsky, Russia, is probably of either Alan-Sarmatian, Kangju, or Kushan origin.
These would probably evolve into the helmets found at Chudjaky Grave 88, being of the Hunnic period (2nd Row, left, with the weird arrangement that makes it look like it has a Gjermundbu-style visor, but it actually doesn't those are fragments of the neckguard).
And going once more back to this image and the Galerius Arch, the Suvorovo Grave 30 (Top middle) helmet looks weirdly like the helmets in the 2nd row of the Galerius Arch. I think the presence of these helmets on the Galerius arch can be explained by Pontic/Caucasian recruits.
Galerius' reign in the late 3rd century was before the beginning of the arms factories and during a period when the first ridge helmets were emerging. Roman units from the Pontic-Danubian and Caucasus regions must have been manufacturing local styles in their workshops.
The alternative, of course, remains Roman styles. The problem is that single-piece conical and multi-piece conical Roman helmets are few and far between, and there's questions of whether or not many of them are fakes.
But that pretty much wraps up this thread. Thanks for reading, these arguments are quite common in the reenactment community and another debate on the Deir El Medniah helmet is what spurred me to do all of this.
One more addition: the predecessor to these lamellar helmets (right). A 1st Century Lamellar Helmet from Isakovskij Kurgan 6.
Three*. Sorry. I totally forgot about Deurne.
