Countess Bertha and Arnulf of Valenciennes

[lancast...@gmail.com]

Peter Stewart has suggested that I explain why Léon Vanderkindere's explanation of Bertha's parentage needs to be rejected. This is something I explained in an article which is unfortunately not freely available online, but it is indeed perhaps a good idea to explain some points in short form.

Firstly, for those who do not know, Vanderkindere was a liberal politician who lived in Brussels, and wrote around 1900. His attempts to explain the origins of the various medieval counties around Belgian are still very influential. (In some ways though they continued a much older early modern tradition of trying to connect the medieval dynasties of the low countries in ways which connected them all backwards to the Carolingians, and forwards to the early modern dynasts who (re)united them.) His Wikipedia article gives links to his 2 volume work on "la formation territoriale des principautés belges". https://en.wikipedia.org/wiki/L%C3%A9on_Vanderkindere

Notes on what Vanderkindere says about Bertha.

1. He takes it as certain that she must have been visiting relatives when she died. Whatever we think of the chronicle of Sint Truiden, it is the only source, and it gives a quite different reasoning. It says that she was on a pilgrimage in the region when she became ill. It specifies that her son Arnulf had to travel from Flanders. (Note: before she died there was time for people to travel there.) I don't see any reason to reject this, but if we insist on rejecting it there is no other evidence which gives an alternative account. The only thing we can say in favour of this position is that she did apparently have some sources of income in the area around Sint-Truiden (tithes in Brustem, a forest in Melveren). However it is equally obvious that she had lands very far away. So it seems prudent to assume that she may have had lands etc scattered over a very large region.

2. Concerning her parentage and relatives, Vanderkindere took it as certain not only that her main relatives lived around Sint-Truiden, but also that she was a relative of Bishop Balderic II of Liège, who would not yet have been born at this stage. This is possible because Balderic's "biography" (the Vita Balderici) makes it clear that Balderic II was a close relative of her son Count Arnulf. But how does Vanderkindere exclude the possibility that Arnulf was related to Balderic via his unknown father? Vanderkindere took it as given that the term "cognatio" had to mean that the two men were related through a female line on Arnulf's side, although he believed it was through Balderic's father. With this rather questionable line of reasoning, which is handled in a few asides, Vanderkindere "proved" that Bertha was the sister of Balderic II's father. This does not even fit well with his other proposals because he believed Balderic II's father was a young boy at this time when Bertha was a lady with a grown-up son.

It would take a lot longer post to explain why historians accepted this (at least in print) for more than a century, but I think it has to do with the fact that the assertion is only one in a much bigger web of assertions, which were presented in a complex way by later defenders of the proposal such as Jean Baerten. Also, from conversations I understand that many historians could see a problem needed unravelling, but such genealogical asides are not always central to the topics they are working on. Finally, in order to write about this topic you are almost forced to write a relatively long negative article which only disproves things, but can't offer any alternative scenario.

[Peter Stewart]

On 19-Mar-23 9:58 PM, lancast...@gmail.com wrote:
> Peter Stewart has suggested that I explain why Léon Vanderkindere's explanation of Bertha's parentage needs to be rejected. This is something I explained in an article which is unfortunately not freely available online, but it is indeed perhaps a good idea to explain some points in short form.
>
> Firstly, for those who do not know, Vanderkindere was a liberal politician who lived in Brussels, and wrote around 1900. His attempts to explain the origins of the various medieval counties around Belgian are still very influential. (In some ways though they continued a much older early modern tradition of trying to connect the medieval dynasties of the low countries in ways which connected them all backwards to the Carolingians, and forwards to the early modern dynasts who (re)united them.) His Wikipedia article gives links to his 2 volume work on "la formation territoriale des principautés belges". https://en.wikipedia.org/wiki/L%C3%A9on_Vanderkindere
>
> Notes on what Vanderkindere says about Bertha.
>
> 1. He takes it as certain that she must have been visiting relatives when she died. Whatever we think of the chronicle of Sint Truiden, it is the only source, and it gives a quite different reasoning. It says that she was on a pilgrimage in the region when she became ill. It specifies that her son Arnulf had to travel from Flanders. (Note: before she died there was time for people to travel there.) I don't see any reason to reject this, but if we insist on rejecting it there is no other evidence which gives an alternative account. The only thing we can say in favour of this position is that she did apparently have some sources of income in the area around Sint-Truiden (tithes in Brustem, a forest in Melveren). However it is equally obvious that she had lands very far away. So it seems prudent to assume that she may have had lands etc scattered over a very large region.
>
> 2. Concerning her parentage and relatives, Vanderkindere took it as certain not only that her main relatives lived around Sint-Truiden, but also that she was a relative of Bishop Balderic II of Liège, who would not yet have been born at this stage. This is possible because Balderic's "biography" (the Vita Balderici) makes it clear that Balderic II was a close relative of her son Count Arnulf. But how does Vanderkindere exclude the possibility that Arnulf was related to Balderic via his unknown father? Vanderkindere took it as given that the term "cognatio" had to mean that the two men were related through a female line on Arnulf's side, although he believed it was through Balderic's father. With this rather questionable line of reasoning, which is handled in a few asides, Vanderkindere "proved" that Bertha was the sister of Balderic II's father. This does not even fit well with his other proposals because he believed Balderic II's father was a young boy at this time when Bertha was a lady with a grown-up son.

Although it can't be certain that Bertha was visiting relatives - the
Sint-Truiden chronicle says that she came to the abbey to pay her
respects to the saint buried there while returning from a pilgrimage to
Aachen - she was evidently travelling from east to west and probably
making her way home towards Valenciennes when she fell ill.

Vanderkindere may not have been far astray in assuming that the close
relationship between her son Arnulf and Balderic II of Liège came
through her. This is not only mentioned in the Vita as noted above but
also in a charter of Balderic himself, dated 1015 ("Defuncto igitur
Arnulfo comite de Valencines, consanguineo meo"). According to the
obituary of Saint-Lambert, Bertha's son Arnulf died on 23 October having
donated "Viusaz" (presumably Visé) to the cathedral. A 12th-century copy
of Usuard's martyrology in Cambrai places the death of Count Arnulf of
Valenciennes under 22 October ("Obiit Arnulfus comes Valentianensis"),
so there is no great difficulty over his identification in the
Saint-Lambert record on the following day. The Saint-Lambert obituary
also records the death of Countess Bertha on 30 October noting that she
had donated Crehen - this is a couple of kms south-west of Hannut,
around 44 kms west of Liège and around 32 kms south-west of Borgloon
held by Balderic's family.

Peter Stewart



--
This email has been checked for viruses by AVG antivirus software.
www.avg.com

[lancast...@gmail.com]

On Monday, March 20, 2023 at 2:45:31 AM UTC+1, Peter Stewart wrote:

Thanks Peter. I see that I scanned that page from Parisse's edition of the necrology but perhaps apparently never really processed it. Crehen is certainly in the right area, and as you note this is definitely a countess. Why the dates differ is a mystery, but quite a common thing in medieval necrologies. Arnulf's son is also mentioned in that necrology by the way.

I agree that it is quite possible and even likely that Bertha is a close relative of Balderic II. Whether she shared his apparent relationships with Dirk III of Holland and Lambert of Leuven (both of whom probably had the right type of royal ancestry) would be more uncertain of course. In the context of the argumentation of the argumentation of Vanderkindere and Baerten, what I think we can not do is to take these possibilities, add them together with others, and use them all as the foundation of any proof of another relationship altogether.

[Peter Stewart]

On 20-Mar-23 4:48 PM, lancast...@gmail.com wrote:
> On Monday, March 20, 2023 at 2:45:31 AM UTC+1, Peter Stewart wrote:
>
> Thanks Peter. I see that I scanned that page from Parisse's edition of the necrology but perhaps apparently never really processed it. Crehen is certainly in the right area, and as you note this is definitely a countess. Why the dates differ is a mystery, but quite a common thing in medieval necrologies. Arnulf's son is also mentioned in that necrology by the way.

It's interesting that the son Adalbert is also titled count and credited
with the same donation as Arnulf - in the father's case spelled "Viusaz"
and in the son's "Viozaz", that I assume both meant Visé. Perhaps
Adalbert was count of Cambrai and Arnulf of Valenciennes, or just his
father's associate count when Arnulf must have been an old man by the
early 11th century (this had happened before in the region without age
being the cause, for example in Flanders with Adalulf the younger
brother of Arnulf I).

Thierry Stasser followed Vanderkindere in adducing the name Adalbert as
evidence that Arnulf's wife who donated property in the Darnau was a
daughter of the count there and a sister of Adalbert I of Namur. Maybe
she was, but the name Adalbert also occurs in the Vermandois family
around this time who had a different Carolingian ancestry from the Namur
lineage.

[mike davis]

You are both obviously more conversant with the sources on these people
than I am, so excuse me if I am blundering in here, but doesnt the St.Truid chronicle
or charter [I'm not sure which] confuse Arnulf and Berta with the counts of
Flanders? There do seem to be many counts and nobles with the name Arnulf
in the sources of the late 10th century. Working out which is which is certainly
a minefield.

I read on wiki that Arnulf lost Valenciennes to Baldwin V 1006, perhaps this
was only temporary as Tanner in her book on Boulogne dates this to 1013,
and that the Emperor gave Cambresis to the Bishop of Cambrai 1007. Then
Arnulf gave Vise to the Bishop to buy his aide to recover his lands, so he
seems to have been in a rather weak position when he died. AIUI the marches
of valenciennes & Eename were erected as a barrier to the Frankish kings
invading Lorraine like Lothar had done. By this time the threat was perhaps over
and they were no longer needed.

Lastly as these threads on Richilde seem to be inclining towards Arnulf of
Valenciennes, I saw a line of descent repeated on many websites like geni.com,
which have the following for Arnulf of Valenciennes:

1 Charles the Bald m Ermentrude

2 Judith m Baldwin I of Flanders d879

3 Raoul [Rodulf] Count of Cambrai k 896

4 dau [sometimes called Judith] m Isaac Count of Cambrai 908-46

5 Arnulf I, son of Isaac 941, Count [of Cambrai?] 960 m Berta of Batavie d967

6 Arnulf II the Younger Ct of Cambrai 979 Markgraf & Ct of Valenciennes 983 d 1011or12 m Letgarde of Namur

The only worthwhile reference was ES II,5 which I take to mean the Europaische Studien series.

Mike

[Hans Vogels]

Op maandag 20 maart 2023 om 19:33:50 UTC+1 schreef mike davis:

Hello Mike,

ES refers to Europäische Stammtafeln. See:
https://johnblythedobson.org/GFA/ES/contents.html

With regards,
Hans Vogels

[lancast...@gmail.com]

Hi Mike
No problem!
Quick notes:

1. Just as an FYI, I think a good English version might for the saint and founder might be St. Trudo. Concerning European placenames I follow the 21st habit of using local spellings, mostly, but not always. In Dutch the town where the abbey was is Sint-Truiden. I live a short drive away. The local dutch dialect has been immortalized in the Oscar-winning film "Rundskop" (Bullhead).

2. You are correct to observe that there is a concern about whether the chronicle of the abbey there was making a typical type of genealogists mistake by calling Bertha and Arnulf "Flemish". Peter is arguing that to a certain point, but in reality we are not far apart (I think). Certainly the chronicle was wrong if it thought Arnulf was "the" count of Flanders. But he clearly was "a" count of Flanders, at least using the term in the way it was used at that time. You make me think of an interesting scenario though. Would the daughters of "the" counts of Flanders be called countesses of Flanders?? Yes they would, and Bertha may well have been a daughter of a count of Flanders.

3. Whether we can equate Arnulf of Cambrai and Arnulf of Valenciennes is not perfectly clear. It is a perfectly reasonable proposal. It is commonly treated as if certain. Peter prefers not to do so, and that is perfectly correct.

4. Unfortunately I don't think Peter or I are very confident about Richilde being a relative of Arnulf and Bertha. We don't seem to have found a strong reason to prefer that option. There are too many options.

5. Concerning the genealogy you mention this is indeed a gorilla in the room for Peter and myself. I think both of us are saying that this typical proposal is very speculative at almost every step. We do not know who Bertha was married to. I think we can simply reject the idea that she was "de Batavie" (the Vanderkindere proposal). Who knows. Maybe she was a daughter of Isaac.

Andrew

[mike davis]

Sorry, yes i seem to have combined english and dutch and got it both wrong.

>
> 2. You are correct to observe that there is a concern about whether the chronicle of the abbey there was making a typical type of genealogists mistake by calling Bertha and Arnulf "Flemish". Peter is arguing that to a certain point, but in reality we are not far apart (I think). Certainly the chronicle was wrong if it thought Arnulf was "the" count of Flanders. But he clearly was "a" count of Flanders, at least using the term in the way it was used at that time. You make me think of an interesting scenario though. Would the daughters of "the" counts of Flanders be called countesses of Flanders?? Yes they would, and Bertha may well have been a daughter of a count of Flanders.
>

So calling them count or countess of Flanders is not a title but a more general geographic term? I can
see that could be the case, but my initial feeling was that the writer thought he was Arnulf the Old.
But I havnt read the rest of the chronicle to see if they are treated as separate persons.

> 3. Whether we can equate Arnulf of Cambrai and Arnulf of Valenciennes is not perfectly clear. It is a perfectly reasonable proposal. It is commonly treated as if certain. Peter prefers not to do so, and that is perfectly correct.
>

Most historians seem to have combined the two, and I also tend to follow that view, but there are plenty of
Arnulfs to choose from at that time so the defender of Cambrai in 979 could be different from the
markgrave of the 983, if those dates are correct.

> 4. Unfortunately I don't think Peter or I are very confident about Richilde being a relative of Arnulf and Bertha. We don't seem to have found a strong reason to prefer that option. There are too many options.
>

Yes I think you are right. There doesnt seem as yet any hard evidence.

> 5. Concerning the genealogy you mention this is indeed a gorilla in the room for Peter and myself. I think both of us are saying that this typical proposal is very speculative at almost every step. We do not know who Bertha was married to. I think we can simply reject the idea that she was "de Batavie" (the Vanderkindere proposal). Who knows. Maybe she was a daughter of Isaac.

I agree. My purpose in posting this was in case people saw these threads and thought they knew
the answers from these descents that are freely available to anyone with a net connection.

Mike

[lancast...@gmail.com]

Mike you asked, "So calling them count or countess of Flanders is not a title but a more general geographic term?" It was a term which was used in different ways, but in this period it was generally not geographical. "The" count of Flanders was of course a case like the ones we expect from later records. He was a count connected to a geographically straightforward territory which was also a county. But the term still very often had a meaning reflecting its Roman origins, signifying office and status, but not necessarily an single stable and contiguous geographical territory called a county. A "comitatus" could be a jurisdiction, and such jurisdictions might have covered specific villas in certain region, but they were not normally a clear way of describing a geographical entity. In this period, when imperial records want to describe a territorial entity they used the terms pagus or gau. There are many records which carefully mention both the pagus which a villa is in, as well as who held the comital jurisdiction over the place. There was clearly no one-to-one correspondence that the users of such records could rely on.

[Peter Stewart]

I don't follow your reasoning for 'he clearly was "a" count of Flanders,
at least using the term in the way it was used at that time' - the time
in question was the late-14th century, as we have no evidence that the
terms used then were taken verbatim from any earlier source. Where do
you find late-14th century references to "comes Flandrensis" or variants
of this meaning "a" count in the region of Flanders rather than "the"
count of Flanders? My opinion is that the Sint-Triuden writer wrongly
identified Bertha as a widowed countess of Flanders and her son Arnulf
as the successor of her deceased husband - but this hardly matters
anyway since it does not negate the historicity of Bertha's visit to
Sint-Truiden on her way home from Aachen, or her illness and death on
the journey, or Arnulf's donation according to her dying wishes.

> 3. Whether we can equate Arnulf of Cambrai and Arnulf of Valenciennes is not perfectly clear. It is a perfectly reasonable proposal. It is commonly treated as if certain. Peter prefers not to do so, and that is perfectly correct.

I don't recall making a comment on this but maybe I've forgotten
something - I thought I had said (or should have done) that there is a
limit to how many contemporaneous counts named Arnulf can be reasonably
postulated, and that identifying the count of Cambrai with the count of
Valenciennes seems sensible enough to me.

> 4. Unfortunately I don't think Peter or I are very confident about Richilde being a relative of Arnulf and Bertha. We don't seem to have found a strong reason to prefer that option. There are too many options.

There are several alternatives for Richilde and Herman between them -
one or other, or perhaps both - having some hereditary claim to
Valenciennes. However, a castellan whose successor was named Isaac had
possession of Valenciennes evidently before (or if not very soon after)
the countship/margraviate was taken up by the Hainaut couple. It is not
unlikely than this man, Hugo, represented the kindred paid off in order
for the fuzzy inheritance to pass into her (or her husband's) hands. I
will get round to this later, as energy allows.

> 5. Concerning the genealogy you mention this is indeed a gorilla in the room for Peter and myself. I think both of us are saying that this typical proposal is very speculative at almost every step. We do not know who Bertha was married to. I think we can simply reject the idea that she was "de Batavie" (the Vanderkindere proposal). Who knows. Maybe she was a daughter of Isaac.

That is a plausible suggestion - if Isaac was succeeded by his
son-in-law Amulric, as some historians suppose, and Amulric was
subsequently divorced from Isaac's daughter for consanguinity, then
Valenciennes may have gone after him to Bertha's husband as another
son-in-law of Isaac. But this is open to other equally plausible
alternative scenarios.

[lancast...@gmail.com]

Thanks for those clarifications Peter. Just in answer to your first remark. I assume that the 14th century writer was basing himself upon sources we no longer have. As he is the only source we have for some events centuries early, this seems to be a difficult conclusion to avoid. So it was therefore possible that it was his source who called Bertha a Flemish countess.
As I think I mentioned somewhere, it might also be relevant to consider that during the period we are talking about the Karlings were imposing themselves upon the southern part of "Flanders" near Lens where Bertha and her son apparently had some sort of foothold. So it was probably not under "the" counts of Flanders.
Whether our 14th century reporter understood all this is another matter. (Although I don't want to disparage them. They seem to have been quite the antiquarian, with an interest in trying to connect the dots in times long past.)

[Peter Stewart]

The trouble with assuming an earlier source for the territorial
designation is that whenever this was added to the titles countess and
count it was inaccurate and almost certainly not in the original
10th-century charter (where such an identifying qualification to a
comital title would have been very rare in the text and unusual even in
a list of subscriptions).

Also, any contemporary of Bertha and Arnulf would not have identified a
countess and count of Valenciennes as belonging to the region of
Flanders - in their time these were understood as two distinct
margraviates in different kingdoms. It would be somewhat like calling a
US senator from Vermont a senator from the Quebec region.

Peter Stewart

[lancast...@gmail.com]

I agree. The 14th century writer's source may have simply indicated that these were people of comital status from the general direction of Flanders. I was not actually proposing any link to Valenciennes itself in the hypothetical Sint Truiden records about Bertha. In fact, calling them Flemish may simply have been based upon the fact that the biggest grant they made was Provin (between Lille and Lens). We know from other records that Arnulf must have had more (perhaps all?) of the small pagus there which gets called Carembault, Carembaut, Caribant etc. In later generations it was the Count of Flanders who confirmed the grant, so that's a good reference for saying these counts were based in "Flanders". But Caribant is almost 50km from Valenciennes, and there was an "international" border, the Scheldt, between them. So I see no reason to suggest that the Sint Truiden monk would have connected this family with Valenciennes, and in general I don't think Valenciennes would have been seen as part of Flanders.

Modern historians and antiquarians found it difficult to prove that Arnulf of Valenciennes was Arnulf the son of Bertha. (Bas Aarts was still not convinced in his relatively recent articles. I think I may have convinced him now.) To make the link requires us to look at the Ghent grants which connect both people to Caribant.

Just to imagine possibilities, Bertha may have been a relative to the last Carolingians (in France, on one side of the Scheldt), whose husband was a count of the Hainaut/Cambrai area (in the proto-empire, on the other side). The couple may have benefited from the Carolingian forays into the south of Flanders. It was a tricky period in this region. Despite appearing in records in Flemish Ghent, her son Arnulf fought on the imperial side with Godfrey the Captive against the Carolingian-backed Reginars. But then a Carolingian, Charles, was given control of the Duchy, giving his friends the Reginars a new foothold in Mons and Louvain. And at the end of his life he faced invasion from the new ally of the Reginars, the count of Flanders. The Reginars and/or counts of Flanders may possibly have been kinsmen or allies of Arnulf's parents, but we know that wouldn't stop this type of conflict.

[Peter Stewart]

It can't very well have been, since Flanders was part of the West
Frankish kingdom while Valenciennes belonged to the German empire. The
fact that a chronicler in the late-14th century assumed a 10th-century
count named Arnulf to have been the ruler of Flanders has no real
bearing on this or on the source of his information, which was quite
probably his abbey's counterpart of the original 967 charter naming
Arnulf with the title count but (in common with the vast majority of
such documents from the time) no territorial designation at all.

> Modern historians and antiquarians found it difficult to prove that Arnulf of Valenciennes was Arnulf the son of Bertha. (Bas Aarts was still not convinced in his relatively recent articles. I think I may have convinced him now.) To make the link requires us to look at the Ghent grants which connect both people to Caribant.

The questions of how many Arnulfs there may have been in the
Valenciennes lineage and of how many other counts named Arnulf there
were in the penumbra of Flanders during the time of Bertha's son cannot
be definitively resolved on the available evidence.

On the first point, Vanderkindere made a slip over the comital title he
wrongly thought had been accorded to Isaac of Cambrai's son Arnulf in a
charter of the count of Flanders dated 5 May 941. He then supposed
(fairly reasonably, even disregarding his error) that this Arnulf son of
Isaac cannot have been the same man as Bertha's son who lived until the
second decade of the 11th century, so that there must have been two
counts named Arnulf in succession after Isaac, the second of whom was
Bertha's son. Jan Dhondt disputed this on the basis that charters were
sometimes subscribed by infants and that Isaac's son may have been a
young child in May 941 - this is true in general but implausible in the
particular instance, since the charter was not an act of Isaac regarding
property in which his infant son may have had any future hereditary
interest, and the charter was issued in Ghent where Isaac is hardly
likely to have travelled from Cambrai with an infant in tow. However, if
his son Arnulf was around 15 at the time of subscribing the 941 charter
he would have been in his mid-80s when dying in 1011/12, and this is not
so exceptional as to require an intervening generation. In other words,
for all we can prove Bertha may have been Isaac's wife, daughter or
daughter-in-law. Isaac's son (or perhaps grandson) Arnulf had lost his
own son and (at least briefly) possession of Valenciennes by the end of
what may have been a long life - and for that matter his heirless
debility in old age may have been a pretext for the count of Flanders to
take over Valenciennes in 1006.

As to how many other counts named Arnulf there may have been, this
cannot be answered with any certainty before (or after) 962, when Arnulf
the son of Adalolf was installed as count of Boulogne & Ternois in a
peace settlement with his paternal uncle Arnulf I of Flanders. Adalolf
had died in 933, and the elder of his two sons (both presumably underage
at the time) may have been called count from then on although his
namesake uncle took over Boulogne and Ternois along with the lay abbacy
of Saint-Bertin until 962. We don't know where Arnulf of Boulogne was in
this long interval, but it is more probable that he lived at the comital
court in Flanders rather than being let loose for 29 years. Adalolf's
sons had the means to oppose their uncle by the early 960s, when the
younger brother was killed and the king intervened to establish the
elder as count in Boulogne. He may have been acknowledged by his uncle
as a titular count from 933 onwards.

There were two counts named Arnulf occurring together as subscribers in
the (pseudo?-original) charter of Balduin IV of Flanders dated 1 April
988, just after the death of Arnulf II of Flanders. One of these is
sometimes thought to have been Arnulf of West Friesland, titled as count
of Ghent although his father was still living and subscribed the same
charter. This identification is further complicated by his having a
close relative - perhaps a maternal half-brother - also named Arnulf,
son of Hilduin, who was possibly count (or son of a count) of Tournai
and/or Montdidier. Fraser McNair in 2017 thought that these two Arnulfs
were both sons by different husbands of Hildegard whom he thought to
have been the sister of Arnulf I of Flanders, or perhaps his daughter.
She can't very plausibly have been his sister, because this would make
her daughter of parents who married in the 880s whereas she herself
married and had offspring in the 940s.

I think it likely that the two counts Arnulf in the 988 charter were
Bertha's son of Cambrai & Valenciennes and either Hildegard's son Arnulf
of West Friesland or Adalolf's grandson Arnulf II (Ernicule) of Boulogne
- however, if the latter was still living there is a further difficulty
in identifying a count Balduin who also subscribed the charter, who may
have been his successor. The chances of yet another count Arnulf cannot
be eliminated, but without additional evidence I can't see a solid basis
for ascribing one to Valenciennes and another to Cambrai.

Peter Stewart