Spell modifier capsFollow

In my travels, I have discoverred the following. There is a cap on spell haste at 50%. There is a cap on duration extensions at 50%. If you have any other information on modifiers that are capped, please post.

Duration Extension Details
I just got an aug for 1450 in north ro. It is the one that has the Extended Blessing IV effect on it. Someone had told me that it stacks with Extended Enhancement IV, so I jumped at it. I timed short and long duration spells in three configurations. In all cases, I have Spell Casting Reinforcement Mastery, which gives a 50% extension to buff duration.

Extended Blessing IV equipped
Extended Enhancement IV equipped
Both Equipped
Neither Equipped

For the short duration spell, I used Feedback, and for the longer, I used Speed of Vallon. My findings were consistent. For all buffs in all configurations, the items made no difference. All buffs were extended by a consistent 48 to 51%, which is close enough to 50% to be covered by my margin of error. I have concluded, based on this data, that buffs cannot be extended by more than 50%.

Buff Haste
This involves how long it takes for buffs to cast. All buffs with a casting time over 3 seconds are affected by this line of spells, items, and aa skills.

I was planning on getting my buff haste to around 90% when I "tested" this out. I got spell casting deftness 3 (25% buff haste) which got me to about a 45% casting time reduction with a Quick Buff IV focus item. I then proceeded to get Quick Buff 3 (50% buff haste - Chanter class skill aa). After spending the 18 aa's, my buff casting haste was 50%. Please note that Quick Buff 3 by itself gets you 50% buff haste and does not require spell casting deftness as a prerequisite.

I then tested buff haste with three other configurations, although I knew what the outcome would be before I measured.

Equipped Buff Haste IV item
Blessing of Replenishment buff
Item and Buff

In all cases, casting times were reduced by exactly 50%, leading me to believe that there is a hard cap at this point. Other data would be welcome if you have it.